Identification document and related methods

[0001] The present application claims the benefit of U.S. Provisional Patent Application Nos. 60/418,762, filed Oct. 15, 2002, 60/421,254, filed Oct. 25, 2002, 60/494,709, filed Aug. 12, 2003, 60/495,236, filed Aug. 13, 2003 and 60/495,373, filed Aug. 14, 2003. Each of these patent documents is herein incorporated by reference.

[0002] This application is also related to assignee's U.S. patent application Ser. No. 09/945,244, filed Aug. 31, 2001 (published as 20020057823 A1), Ser. No. 09/503,881 (now U.S. Pat. No. 6,614,914), Ser. No. 09/452,023 (now U.S. Pat. No. 6,408,082), Ser. No. 10/366,541, filed Feb. 12, 2003 (published as US 2003-0179903 A1), Ser. No. 09/433,104, filed Nov. 3, 1999 (allowed), Ser. No. 10/032,282, filed Dec. 20, 2001 (published as US 2002-0114491 A1), and Ser. No. 09/498,223, filed Feb. 3, 2000 (now U.S. Pat. No. 6,574,350), and U.S. Pat. No. 6,389,151. The Ser. No. 10/366,541 application is a grandchild of assignee's U.S. Pat. No. 5,841,886. This application is also related to assignee's concurrently filed U.S. patent application titled “DIGITAL WATERMARKING FOR IDENTIFICATION DOCUMENTS” (Attorney Docket No. P0869—Inventors J. Scott Carr, et al.). Each of these patent documents is herein incorporated by reference.

[0003] The present invention relates to authenticating and securing identification documents through digital watermarking.

[0004] The use of identification documents is pervasive. Identification documents are used on a daily basis—to prove identity, to verify age, to access a secure area, to evidence driving privileges, to cash a check, and so on. Airplane passengers are required to show an identification document during check in, and sometimes at security screening and prior to boarding their flight. We also live in an ever-evolving cashless society. Identification documents are used to make payments, access an automated teller machine (ATM), debit an account, or make a payment, etc. Many industries require that their employees carry photo ID on the job.

[0005] For the purposes of this disclosure, identification documents are broadly defined and may include, e.g., credit cards, bank cards, phone cards, passports, driver's licenses, network access cards, employee badges, debit cards, security cards, visas, immigration documentation, national ID cards, citizenship cards, social security cards, security badges, certificates, identification cards or documents, voter registration cards, police ID cards, border crossing cards, legal instruments or documentation, security clearance badges and cards, gun permits, gift certificates or cards, labels or product packaging, membership cards or badges, etc., etc. Also, the terms “document,” “card,” and “documentation” are used interchangeably throughout this patent document. Identification documents are also sometimes interchangeably referred to as “security documents,” “ID documents,” “photo-IDs” and “photo ID documents”.

[0006] With reference to FIG. 1, an identification document 10 includes a “card-shaped” substrate 21, historically made from a material such as paper or plastic, but now even made from synthetics such as Teslin®. (Teslin® is available from PPG Industries, One PPG Place, Pittsburgh, Pa. 15272 USA). Identification document 10 includes a photograph 14 and various data 12, e.g., such as textual information, graphics, a screened-back or hidden image, bar codes, biometric information (e.g., a fingerprint), text information (e.g., name, address, birth date, ID number, etc.), or the like. Of course both sides of substrate 21 can receive printing or engraving. Other suitably interchangeable features and materials are found, e.g., in assignee's U.S. patent application Ser. No. 10/330,032, filed Dec. 24, 2002 (published as US 2003-0173406 A1), which is herein incorporated by reference.

[0007] The printed substrate 21 is usually laminated. The laminate typically includes a plastic, polyester or polycarbonate-based top sheet 23 and bottom sheet 25 that respectively overlay the top and bottom of the substrate 21. Heat and/or adhesives and pressure are used to bond the laminate sheets 23 and 25 with the substrate 21. Or a laminate can include a pouch into which the substrate 21 slips. Again, heat and/or adhesives and pressure are used to bond the substrate 21 with a pouch laminate. The laminates provide a protective covering for the printed substrate and provide a level of protection against unauthorized tampering. (For example, a laminate would have to be removed to alter the printed information and then subsequently replaced after the alteration.). A laminate layer 23 or 25 may optionally carry information like a card bearer's signature or security features.

[0008] In some implementations, information may also be optically or magnetically stored on recording media (e.g., magnetic stripe 27, or optical memory or electronic circuitry—not shown in FIG. 1) carried by the laminate 25. Of course the recording media can be alternatively carried by substrate 21 or laminate 23.

[0009] We note that the present invention encompasses ID documents including more or less features and layers than are illustrated in FIG. 1. Additional features may include graphics printed in dual color optically variable ink, microprinting text, so-called “ghost images,” information and images printed in UV inks, and security features such as those disclosed in, e.g., assignee's U.S. patent application Ser. No. 10/170,223 (published as US 2003-0031340 A1), which is herein incorporated by reference.

[0010] Identification documents can also include information such as a bar code (e.g., which may contain information specific to the person whose image appears in the photographic image, and/or information that is the same from ID document to ID document), variable personal information (e.g., such as an address, signature, and/or birth date, biometric information associated with the person whose image appears in the photographic image, e.g., a fingerprint), a magnetic stripe (which, for example, can be on the a side of the ID document that is opposite a side with a photographic image), and various security features (e.g., a security pattern like a printed pattern comprising a tightly printed pattern of finely divided printed and unprinted areas in close proximity to each other, such as a fine-line printed security pattern as is used in the printing of banknote paper, stock certificates, and the like). Of course, an identification document can include more or less features.

[0011] Another example of an identification document is one including a core layer (which can be pre-printed), such as a light-colored, opaque material, e.g., TESLIN, which is available from PPG Industries) or polyvinyl chloride (PVC) material. The core can be laminated with a transparent material, such as clear PVC to form a so-called “card blank”. Information, such as variable personal information (e.g., photographic information, address, name, document number, etc.), is printed on the card blank using a method such as Dye Diffusion Thermal Transfer (“D2T2”) printing (e.g., as described in commonly assigned U.S. Pat. No. 6,066,594, which is herein incorporated by reference), laser or ink jet printing, offset printing, etc. The information can, for example, comprise an indicium or indicia, such as the invariant or nonvarying information common to a large number of identification documents, for example the name and logo of the organization issuing the documents. Any known process capable of forming the indicium may be used to form the information.

[0012] To protect the information that is printed, an additional layer of transparent overlaminate can be coupled to the card blank and printed information, as is known by those skilled in the art. Illustrative examples of usable materials for overlaminates include biaxially oriented polyester or other optically clear durable plastic film.

[0013] Of course, there are many other identification documents that include different structures, features and materials. These other identification documents can be suitably interchanged with the identification documents described herein. The inventive digital watermarking techniques disclosed herein will similarly benefit these other documents as well.

[0014] An identification document may include a digital watermark. Digital watermarking is a process for modifying physical or electronic media to embed a machine-readable code into the media. The media may be modified such that the embedded code is imperceptible or nearly imperceptible to the user, yet may be detected through an automated detection process. In some of our preferred embodiments, an identification document includes two or more digital watermarks.

[0015] Digital watermarking systems typically have two primary components: an encoder that embeds the digital watermark in a host media signal, and a decoder that detects and reads the embedded digital watermark from a signal suspected of containing a digital watermark (a suspect signal). The encoder embeds a digital watermark by altering the host media signal. The alterations usually take the form of altered signal values, such as slightly changed pixel values, luminance, colors, changed DCT coefficients, altered signal values or selective placement or signal tweaks, etc. However, a watermark can also be manifested in other ways, such as changes in the surface microtopology of a medium, localized chemical changes (e.g. in photographic emulsions), localized variations in optical density, localized changes in luminescence, etc. The surface texture of an object may be altered to create a watermark pattern. This may be accomplished by manufacturing an object in a manner that creates a textured surface or by applying material to the surface (e.g., an invisible film or ink) in a subsequent process. The watermark reading component analyzes content to detect whether a watermark pattern is present. In applications where the watermark encodes information, the reading component extracts this information from the detected watermark. The reading component analyzes a suspect signal to detect whether a digital watermark is present. The reading component can be hosted on a wide variety of units ranging from tethered or wireless reader devices, conventional personal computers, network servers, cell phones including cameras, to fully mobile readers with built-in displays. Image data corresponding to a watermarked surface of an identification document is read and decoded by this reader to obtain a watermark's information or “payload”.

[0016] Several particular digital watermarking techniques have been developed. The reader is presumed to be familiar with the literature in this field. Some techniques for embedding and detecting imperceptible watermarks in media signals are detailed in assignee's U.S. Pat. No. 6,614,914, U.S. Pat. No. 6,122,403 and PCT patent application PCT/US02/20832 (published in English as WO 03/005291), which are each herein incorporated by reference.

[0017] In assignee's U.S. Pat. No. 5,841,886 techniques and methods are disclosed to detect alteration of photo ID documents, and to generally enhance the confidence and security of such systems. In this regard, reference is made to FIG. 2, which depicts a photo-ID card or document 1000 which may be, for example, a passport or visa, driver's license, credit card, government employee identification, or a private industry identification badge. For convenience, such photograph-based identification documents will be collectively referred to as photo ID documents.

[0018] The photo ID document includes a photograph 1010 that is attached to the document 1000. Printed, human-readable information 1012 is incorporated in the document 1000, adjacent to the photograph 1010. Machine-readable information, such as that known as “bar code” may also be included adjacent to the photograph. Generally, the photo ID document is constructed so that tampering with the document (for example, swapping the original photograph with another) should cause noticeable damage to the card. Nevertheless, skilled forgers are able to either alter existing documents or manufacture fraudulent photo ID documents in a manner that is extremely difficult to detect. Security associated with the use of photo ID documents is enhanced by supplementing the photographic image with encoded information (which information may or may not be visually perceptible), thereby facilitating the correlation of the photographic image with other information concerning the person, such as the printed information 1012 appearing on the document 1000.

[0019] A photograph 1010 may be produced from a raw digital image to which is added a master snowy image as described in U.S. Pat. No. 5,841,886. A central network and point-of-sale reading device (which device, in the present embodiment, may be considered as a point-of-entry or point-of-security photo ID reading device), would essentially carry out the same processing as described with that embodiment, including the central network generation of unique numbers to serve as indices to a set of defined orthogonal patterns, the associated dot product operation carried out by the reader, and the comparison with a similar operation carried out by the central network. If the numbers generated from the dot product operation carried out by the reader and the central network match, in this embodiment, the network sends the OK to the reader, indicating a legitimate or unaltered photo ID document.

[0020] It will be appreciated that the information encoded into the photograph may correlate to, or be redundant with, the readable information 1012 appearing on the document. Accordingly, such a document could be authenticated by placing the photo ID document on a scanning system, such as would be available at a passport or visa control point. The local computer, which may be provided with the universal code for extracting the identification information, displays the extracted information on the local computer screen so that the operator is able to confirm the correlation between the encoded information and the readable information 1012 carried on the document. It will be appreciated that the information encoded with the photograph need not necessarily correlate with other information on an identification document. For example, the scanning system may need only to confirm the existence of the identification code so that the user may be provided with a “go” or “no go” indication of whether the photograph has been tampered with. It will also be appreciated that the local computer, using an encrypted digital communications line, could send a packet of information to a central verification facility, which thereafter returns an encrypted “go” or “no go” indication.

[0021] In another implementation of the '886 patent, it is contemplated that the identification code embedded in the photograph may be a robust digital image of biometric data, such as a fingerprint of the card bearer, which image, after scanning and display, may be employed for comparison with the actual fingerprint of the bearer in very high security access points where on-the-spot fingerprint recognition systems (or retinal scans, etc.) are employed.

[0022] We disclose herein methods, systems and identification document to improve or provide alternative techniques for confronting fraudulent use of identification documents. Fraudulent use of identification documents may occur where, for example, an otherwise legitimate identification document is modified such that the original photograph is swapped with that of another person, thereby enabling the other person to masquerade, at least temporarily, under the guise of the original identification document holder.

[0023] Even in the absence of photo swapping or alteration, it is oftentimes difficult to confirm by inspection that the individual depicted in the photograph of the identification card is indeed the bearer of the card.

[0024] One aspect of this invention provides enhanced security and certainty in the use of photo identification documents.

[0025] Another aspect provides methods and systems for authenticating and securing identification documents using multiple digital watermarks or multiple watermark components.

[0026] According to still another aspect of the present invention, a security document includes a substrate and a first graphic carried by the substrate. The first graphic conveys a photographic image to human viewers thereof, and the first graphic is steganographically encoded to convey first plural bits of data recoverable by computer analysis of the first graphic. The security document also includes a second graphic carried by the substrate. The second graphic conveys a visual image to human viewers thereof, and the second graphic is steganographically encoded to convey second plural bits of data recoverable by computer analysis of said second graphic. The steganographically encoded first plural bits of data and the steganographically encoded second plural bits of data cooperate to verify authenticity of the security document.

[0027] According to still another aspect of the invention, a method is provided to detect swapping of first artwork from a first identification document with second artwork on a second identification document. The swapping results in the first artwork being carried on the second identification document instead of the second artwork. The first artwork includes a first digital watermark embedded therein, and the second identification document includes a second digital watermark embedded in a first region. The method includes: receiving scan data associated with at least a portion of the first artwork and at least a portion of the first region; analyzing the scan data to detect the first digital watermark and the second digital watermark; and comparing the first digital watermark with the second digital watermark to detect swapping of the first artwork with the second artwork.

[0028] Another aspect of the present invention is a security document including a substrate and a first graphic carried by the substrate. The first graphic conveys a photographic image to human viewers thereof, and the first graphic is steganographically encoded to convey first plural bits of digital data recoverable by computer analysis of said first graphic. The security document also includes a second graphic, the second graphic conveying a visual image to human viewers thereof. The security document also includes a detection trigger. The detection trigger serves to indicate a presence of steganographic encoding. The second graphic is steganographically encoded to convey second plural bits of digital data recoverable by computer analysis of said second graphic, wherein the steganographically encoded first plural bits of digital data and the steganographically encoded second plural bits of digital data cooperate to verify authenticity of the security document.

[0029] Still another aspect of the present invention is a method of identifying a first area and a second area of a printed document that are likely to include, respectively, a first digital watermark and a second digital watermark. The first digital watermark includes a first orientation component and the second digital watermark includes a second orientation component. The method includes receiving optically scanned image data that corresponds with at least a portion of the printed document; segmenting the image data into a plurality of image portions; determining an orientation measure relative to a predetermined orientation for each of the image portions; identifying the first area by associating image portions having a first orientation measure; and identifying the second area by associating image portions having a second orientation measure

[0030] Additional features, aspects and advantages of the present invention will become even more apparent with reference to the following detailed description and accompanying drawings.

[0031] FIG. 1 illustrates an identification document.

[0032] FIG. 2 illustrates another identification document.

[0033] FIG. 3A is a diagram illustrating an authenticating method according to one aspect of the present invention; and FIG. 3B is a flow diagram illustrating an authenticating aspect of the FIG. 3A diagram.

[0034] FIG. 4 illustrates an authenticating method according to still another aspect of the present invention.

[0035] FIG. 5 is a flow diagram illustrating a versatile authentication process, which can operate in different authentication modes depending on a watermark detection trigger.

[0036] FIG. 6 illustrates an identification document including two areas, with each area hosting a different digital watermark.

[0037] FIG. 7 illustrates image capture of the FIG. 6 identification document.

[0038] FIGS. 8A-8C illustrate window-based detection.

[0039] FIGS. 9A and 9B illustrate histograms, which classify orientations of different image windows.

[0040] FIGS. 10A-10D illustrate classification of image windows based at least on a watermark orientation component include a translation measure.

[0041] FIG. 11A illustrates two document areas that likely include watermark information.

[0042] FIG. 11B illustrates an isolated window.

[0043] FIG. 12 illustrates a histogram classification of image windows based as least in part on digital watermark strength.

Introduction

[0044] The following detailed description is grouped into four main sections. The grouping is not meant to limit the scope of the present invention, and is merely provided for the reader's convenience. Indeed, we anticipate that disclosed methods, systems and identification documents found under one section heading can be readily combined with methods, systems and identification documents under other section headings. The main sections include: i) Authenticating Identification Documents with Digital Watermarks; ii) Advantages of Watermarking Identification Documents; iii) Detecting Media Areas Likely of Hosting Watermarks; and iv) Introduction to Appendix A.

Authenticating Identification Documents with Digital Watermarks

[0045] First Embodiment—Forensics and Proof of Compliance

[0046] As discussed above with respect to our '886 patent, a photo-ID can be enhanced through embedding data therein. Data can be embedded in, e.g., a graphic or background of the photo-ID. More preferable, however, is to embed the data in a photograph image. The watermark is preferably robust, in that it survives scanning and printing. In some implementations of this embodiment, the watermark is repeated or redundantly embedded throughout an image or document.

[0047] Although sometimes redundantly embedded across an entire document surface, a watermark can be embedded in a particular region of the document if desired (e.g., only in a photograph or a graphic). In an area-based embedding implementation, we can use a mask to identify areas for embedding (or areas that will not receive embedding). The watermark is embedded in identified areas (or is embedded to avoid areas). The watermark can also include a so-called orientation component. An orientation component is helpful in resolving issues of image distortion such as scaling, rotation, translation, etc. In some cases the orientation component comprises a separate digital watermark.

[0048] A digital watermark can be used to provide a machine-readable means to detect fraud due to photo swapping and so-called card “simulation.” The watermark preferably includes a payload or message. The message, once decoded, can be used to provide a visual check for alterations. The decoded message or payload can be displayed, e.g., via a graphical user interface, for use by an inspector or law enforcement officer. For example, if the payload includes the cardholder's date of birth, the payload date of birth can be checked against the corresponding date of birth as printed on the photo-ID. The crosscheck can even be automated. The payload information can include additional information such as driver's license number, jurisdiction code (e.g., code or text indicating which state, agency or company issued the photo-ID), personal information (address, age, sex, weight, height, etc.), driver's insurance information (or a link thereto), unique identifier, etc.

[0049] An advantage of this implementation is that if a watermarked photograph is copied onto another photo-ID document, the watermark provides a telltale sign pointing to the originating card. Once a watermark photograph is determined to be non-compliant with the text or information provided on the document, e.g., the payload data does not correspond with the printed or stored information, the payload can be used to identify the originating document—providing a forensic tracking mechanism. (If the payload includes the originating driver's license number or other unique identifier, such payload information can be used to positively identify the originating document.).

[0050] Another advantage of this implementation is that a machine record for inspecting the photo-ID can be used as a “proof of compliance” or proof of identification inspection. To facilitate such, the watermarked photo-ID is presented to an optical sensor. The optical sensor captures an image of the photo-ID and produces scan data corresponding to such. A watermark detector analyzes the scan data to find and decode the digital watermark embedded in the photo-ID. The detector obtains the payload (e.g., a unique identifier or driver's license number) and a check for the correspondence between the payload and information carried by the card is made. The payload can be stored in a repository or list to evidence that the photo-ID was checked or inspected.

[0051] In an optional implementation of this embodiment, the payload is cross-related to information carried in a magnetic stripe on the card. Validating the card then requires two reads—one an optical read of the watermark on the face of the card, and a swipe of the magnetic stripe through a magnetic-stripe reader. The magnetic stripe information is decoded and compared with the watermark payload. This implementation has the benefit of locking the watermarked information to the magnetic stripe, but it requires an additional read (e.g., for the magnetic stripe) of the card.

[0052] Second Embodiment—Multiple Watermarks

[0053] Additional security is added to an identification document by providing first and second digital watermarks on the identification document. For example, a first digital watermark is embedded in first region of the identification document like in a photograph, artwork, graphic, seal, or image (e.g., ghost image). And a second digital watermark is provided in a second region of the identification card. For example, the second digital watermark is embedded in a background pattern or tint, line-art (see, e.g., assignee's U.S. Pat. No. 6,449,377) or in text, artwork, seal, images or graphics carried by the identification document. The first and second watermarks also may include a so-called orientation component. An orientation component is helpful to resolve image distortion such as rotation, scale and translation. (We note that the second watermark can be embedded using the same or different embedding protocol as the first watermark.). Although not necessary, the first and second watermarks are preferably redundantly embedded or tiled in their respective embedding regions.

[0054] The first digital watermark preferably includes a first payload or first message bits carrying or relating to a first set of information, and the second digital watermark preferably includes a second payload or second message bits carrying or relating to a second set of information. The payloads may include plural-bit payload structure. The payloads can have fixed and variable portions and/or error correction bits. In some cases a payload will include a checksum or error bits to ensure the validity of a payload or a cross-correlation between the watermark payloads. Examples of watermark payloads are shown, e.g., in assignee's U.S. published patent application Ser. No. 10/193,719, filed Jul. 10, 2002 (published as US 2003-0033530 A1) and in U.S. Pat. No. 6,614,914. Each of these patent documents is herein incorporated by reference.

[0055] The first set of information preferably relates to the holder of the identification card (hereafter “cardholder”) and/or to the issuing authority (e.g., state DMV or company) or jurisdiction. For example, the first set of information may include “related information” such as a unique identifier that is associated with the issuer or cardholder, a date of birth, jurisdiction code, identification document number, name, address, physical characteristics (hair color, weight, biometric, sex, etc.), issue date, restrictions (e.g., age restrictions, driving limitations, etc.), a hash (e.g., a reduced-bit representation) of such related information, and/or a combination of the above related information.

[0056] The second set of information, which is carried by or related to the second payload, preferably corresponds with the first set of information. The term “corresponds” in this context is used broadly and may range, e.g., from an exact match to a loose association within a predetermined tolerance.

[0057] In a first implementation, the second set of information corresponds exactly with the first set of information. The sets of information are compared to determine authenticity. In a second implementation, the second set of information includes a subset of the first set of information. The subset is cross-correlated with the first set of information to determine authenticity. In another implementation, the first set of information includes a key to decrypt the second set of information (or vise versa). (We note that an encrypted watermark payload may optionally be decrypted with a key contained in a watermark detector.). In still another implementation, the second set of information includes at least some information that should correspond with information carried by a magnetic stripe or barcode (see a related discussion under the First Embodiment, above). In yet another implementation, the second set of information includes both a subset of the first information and additional information. For example, the subset may include the birth date and document number, while the additional information may correspond with text printed on the document. Or the subset may include simply the document number (or portion thereof), or the bearer's birth date and date of issue. Of course, many other combinations of related information can be provided. For example, the sets of information may be targeted to detect an age or name alteration (e.g., by including age or name information in one or both of the sets of information). In some cases the sets of information includes hashes or reduced bit representations of information pertaining to the cardholder or printed text information. One hash, perhaps a hash carried by the second set of information, may even represent some or all of the first set of information. In still other implementations a document number is distributed between the first and second sets of information. For example, the first set of information includes a first part of the document number and the second set of information includes the remaining portion of the document number. A checksum or error bit can be included with the first or second sets of information to ensure that the document number portions correspond in the expected manner.

[0058] To authenticate an identification document including two digital watermarks, a watermark detector reads both watermarks. The first set of information and the second set of information are retrieved from their respective watermark payloads. (We note that typically only one optical scan is needed to generate scan data, from which both the first and second watermarks are detected from, when the first and second watermarks are provided on the same surface of the identification document.). The first and second sets of information are compared to determine a correspondence. If a correspondence occurs, some or all of the watermark information may be provided to an inspector to facilitate a further check against text alteration. For example, both the date of birth and some data to verify against printed text (e.g., an indication that the 3^rd letter of the first name should be “e”, and the 2^nd letter of the last name should be “t”) can be provided to the inspector. Or, in other implementations, a signal is generated to indicate the authenticity of the identification document.

[0059] FIGS. 3A and 3B are diagrams illustrating another authentication technique for an identification document including digital watermarking. An input device captures an image of an identification document. The identification document includes first and second digital watermarks. The input device conveys data corresponding to a captured image to a watermark reader. The watermark reader can be realized as a programmable computer, which executes software instructions to detect and decode the first and second digital watermarks included in the captured image data. The computer can include a handheld device, laptop, desktop or remote server. (While the input device is illustrated as being tethered to the watermark detector/computer, this aspect of the present invention also contemplates that the input device can wirelessly communicate with the computer or communicate via a network instead. The input device may also be incorporated with the reader, e.g., as in a camera cell phone.). The watermark reader passes decoded watermark information (e.g., payload or message information) to the authenticator. The authenticator can also be realized by software executing on a computer, and in some implementations, the watermark reader includes the authenticator module. The authenticator module determines whether the first and second watermark information corresponds. This determination, in some implementations, involves a crosscheck between a first subset (e.g., birth date) from the first watermark and a first subset from the second watermark (e.g., birth date). In other implementations the authentication determination involves multiple crosschecks between the first and second watermarks. For example, a first subset (e.g., birth date) carried by the first watermark is compared against a first subset (e.g., birth date) of the second watermark, and a second subset (e.g., document number or a portion thereof) of the first watermark is compared against a second subset (e.g., document number or a hash of the document number) of the second digital watermark.

[0060] In an alternative implementation, the authenticator (or watermark reader) passes all or a portion of the watermark information to a computer display (e.g., a computer graphical user interface). Displaying some or all of the watermark information allows an inspector or officer to visually compare the watermark information against information printed on the document. The authenticator outputs an authentication signal to indicate the authentication status of the identification document. In some cases the signal is conveyed to the display. Here, the authentication signal can be a simple pass or fail signal, or a more detailed response indicating the reason for the failure (e.g., one watermark found but the second one is undetectable). In other cases the signal is conveyed to an audio output device (e.g., audio speaker) to audibly signal the authentication status (e.g., a predetermined sound or audio segment is output if authentic, while another predetermined sound or audio segment is output is not authentic). In other implementations a fail signal is used to trigger a secondary process. For example, a fail signal is used to spawn a process to obtain additional data from the identification document. The additional data can be retrieved from optical scan data (e.g., OCR, barcode, or image data for another watermark detection attempt), optical or electronic memory, or manual input. The additional data can be used, e.g., to: i) verify watermark detection failure; ii) authenticate the document through a non-watermarking process; and/or iii) trigger a forensic analysis of the identification document.

[0061] The authenticator module is further discussed with reference to FIG. 3B. The identification document is preferably considered authentic when both the first and second digital watermarks are recovered and when the first and second watermark information (e.g., sets of information) correspond. The document is considered not authentic when either of these criteria is not met. Of course, as discussed above, a watermark detection failure may be used to trigger another process, like a non-watermark based authentication process.

[0062] We note that the second embodiment of this section is generally issuing source and document neutral. This means that a first and second digital watermark is used to verify an identification document regardless of the features provided on the card. Although document neutral, interpretation of the payload bits can vary according to issuer identification. To illustrate, consider a first employee badge issued by a first company and a second employee badge issued by a second company. Each of the first and second employee badges include first and second watermarks embedded therein. All four of the watermarks (i.e., two from the first badge and two from the second badge) have, e.g., a 72-bit payload. The watermarks include a subset of bits (e.g., 4-12 bits) to designate which company issued the badges or to which company the badges are associated with (e.g., an issuer or client code). The bits are located in a predetermined payload location, so that a detector or an application cooperating with the detector can grab the subset of bits and make a determination of who issued the badge. The watermark detector (or a cooperating software application) interprets the bits as designated for the particular issuer or client. For example, the detector or application can query a data structure, branch into a predetermined block of software code, or interrogate a look-up-table to decide what the bits correspond to. (For example, for issuer “Ace,” a private toolmaker, the first 10 bits are the issuer code, the next 32 bits represent a document number, the next 10 bits represent an issue date or birth date and the last 20 bits represent error correction bits or checksums. And, for “Beta,” a government agency, documents associate with Beta include a payload with the first 10 bits representing the issuer code, the next 20 bits represent a document number, the next 20 represent a security clearance, the next 10 represent an issue date and the last 10 represent error correction bits or checksums.). A watermark may similarly carry data (or a signal feature) to indicate that it is a first or second watermark, or an image frame location indicating where a watermark is located is used to distinguish between a first and second watermark. The payloads fields can be designated or interpreted differently for each of the first and second watermarks, even for watermarks issued by the same issuer.

[0063] As an alternative implementation of the second embodiment, a second digital watermark is provided on a different document surface than is the first digital watermark (e.g., the back side of the document). We note that this alternative implementation may require two optical scans to detect both the first and second digital watermarks. This may be less of an issue when the second digital watermark includes information that is used for forensic tracking purposes. For example, the watermark may include information that is tied to the original cardholder. If the second watermark is copied and transferred to a second identification document, the watermark information can be used to trace back to the original cardholder. Similarly, the second watermark may include information pertaining to the issuing location (e.g., which DMV branch office) or original issuing officer.

[0064] A few illustrative combinations are shown below. Of course, these combinations are not meant to limit the scope of this aspect of the invention. Rather, the combinations are provided by way of example only.

[0065] A1. A method to authenticate an identification document, the identification document comprising a first digital watermark and a second digital watermark, said method comprising the steps of:

• • [0066] receiving an optically captured image of at least a portion of the identification document;
  • [0067] analyzing the optically captured image to detect the first and second digital watermarks;
  • [0068] if either of the first or second digital watermarks is not detected, determining that the identification document is not authentic; and
  • [0069] if both of the first and second digital watermarks are detected, comparing the first and second digital watermarks to determine if they correspond.

[0070] A2. The method of combination A1, further comprising the step of displaying through a computer based graphical user interface at least a portion of information carried by either the first or second digital watermark, and comparing the displayed information with text printed on the identification document.

[0071] A3. The method of combination A1, wherein the document comprises text printed thereon, and at least a portion of information carried by the first and second digital watermark corresponds to the printed text.

[0072] A4. The method of combination A1, wherein the first digital watermark includes a first plural bit message, and the second digital watermark includes a second plural bit message.

[0073] A5. The method of combination A4, wherein said comparing of the first and second digital watermarks to determine if they correspond comprises comparing at least a portion of the first message with at least a portion of the second message.

[0074] A6. The method of combination A6, wherein at least one of the first message and the second message includes an issuer code, wherein said combination further comprises interpreting the plural bits of at least one of the first message and the second message according to a predetermined format associated with the issuer code.

[0075] A7. The method of combination A4, wherein the first message comprises a first subset and a second subset, and wherein the second message comprises a third subset and a fourth subset.

[0076] A8. The method of combination A7, wherein said comparing of the first and second digital watermarks to determine if they correspond comprises comparing the first subset with the third subset and comparing the second subset with the fourth subset.

[0077] A9. The method of combination A8, wherein said first subset and third subset comprise redundant information.

[0078] A10. The method of combination A8, wherein said first subset and third subset comprise corresponding information.

[0079] A11. The method of combination A9, wherein the information corresponds through at least one of a cryptographic permutation, a hash and a predetermined key.

[0080] B1. An identification document comprising a first graphic and a second graphic, the first graphic comprising a first digital watermark embedded therein having a first plural-bit payload, the second graphic comprising a second digital watermark embedded therein having a plural-bit payload, the identification document further comprising indicia thereon, wherein the first plural-bit payload comprises a representation of at least a portion of the indicia, and wherein the second plural-bit payload comprises information corresponding to the first plural-bit payload.

[0081] B2. The document of B1, wherein the information comprises the representation.

[0082] B3. The document of B1, wherein the information comprises at least one of a cryptographic permutation of the information, a hash of the representation, and a portion of the representation.

[0083] B4. The document of B1, wherein the representation comprises at least one of a hash of the indicia, a redundant version of the indicia and a subset of the indicia.

[0084] B5. The document of B1, wherein at least one the second payload comprises a checksum to verify the information.

[0085] Third Embodiment—Fragile Watermarks

[0086] A fragile or semi-fragile watermark is provided to any of the first and second embodiments as either a watermark replacement or to augment the above implementations. For example, in the second embodiment, a fragile watermark may be used as either the first or second watermark or as a third watermark component. In some implementations we preferably use our out-of-phase embedding techniques, e.g., as disclosed in PCT/US02/20832 (published in English as WO 03/005291), to embed a fragile watermark. It will be appreciated that a fragile watermark is designed to be lost or to predictably degrade upon certain types of signal processing. A semi-fragile watermark is designed to withstand normal signal processing, but is destroyed or predictably degrades upon malicious attacks.

[0087] The addition of a fragile or semi-fragile watermark adds protection against anticipated fraud scenarios by providing alerts when copies are made. Alteration in conjunction with card copying can be detected from the absence or condition of the fragile watermark.

[0088] Fourth Embodiment—Linking

[0089] In our fourth embodiment we provide a machine-readable link to related information. The machine-readable link is preferably provided via a digital watermark payload or identifier. The identifier can include a unique number that is used to interrogate a database or access a remote resource. In some cases the identifier includes a URL or a code that is used to access an appropriate URL. In a driver's license scenario, a digital watermark includes a link to an insurance database. The database includes data records evidencing that a cardholder does or does not have car insurance. In other cases, the digital watermark includes a link to a DMV database, to allow verification of information printed on the identification document, and perhaps a photograph of the cardholder. The database cardholder can be compared against the person presently holding the card. A “photo swap” can be further detected from comparison of a database photograph with a photograph carried on the card and a visual inspection of the current cardholder. The techniques discussed in assignee's U.S. patent application Ser. No. 09/571,422, filed May 15, 2000, and in U.S. Pat. No. 6,408,331 can be suitable interchanged with this linking aspect of the present invention. These patent documents are herein incorporated by reference.

[0090] Fifth Embodiment—Legacy Detection

[0091] Our fifth embodiment relates generally to handling ID documents including so-called legacy ID documents. The term legacy is used to generally refer to documents, which were previously produced with or without a particular feature like a digital watermark, but which differ in some manner from newly produced documents (e.g., they have or lack the feature). Consider a passport-issuing scenario in which renewal dates are spaced far apart (e.g., 10 years). Unless a total recall of old passports is issued, the population of outstanding passports will include both old passports (e.g., legacy documents) and new passports (e.g., which may include digital watermarking).

[0092] A challenge results for an ID document authentication process that authenticates a population of ID documents including both legacy documents (e.g., with no watermarking) and watermarked documents. To address this challenge we provide methods and systems to determine whether a digital watermark should be present in a document or whether the document was produced prior to the introduction of watermark embedding.

[0093] Assignee's U.S. Published Patent Application No. US 2002-0080994 A1, which is herein incorporated by reference, discusses a situation where a smart card carries information to trigger a watermark reader to look for a digital watermark. The trigger handles a case of legacy cards. In other words, the trigger serves as an indication to a watermark reader to check for a digital watermark. This allows reading new and legacy documents by the same reader.

[0094] We expand these techniques to further help differentiate between watermarked and non-watermarked ID documents, and/or to decide whether to search for a digital watermark on an ID document.

[0095] With reference to FIG. 4, ID document 100 includes a plurality of features including machine-readable features. The machine-readable features may include, e.g., a barcode (e.g., 1 or 2D barcodes) 60, magnetic stripe 27, optical memory (not shown), electronic circuitry (not shown), text for optical character recognition, etc., etc.

[0096] In a first implementation a machine readable feature includes a watermark search or detection trigger. For example, an issue date, or bits corresponding to an issue date, is stored in a magnetic stripe, optical memory, a barcode or other machine-readable indicia. The issued data serves as a “trigger” to evidence whether and ID document should or should not including watermarking.

[0097] With reference to FIG. 4, a reader, e.g., a barcode reader, a magnetic stripe reader, optical memory reader, optical character recognizer (OCR), etc., obtains a watermark trigger (e.g., issue date). Instead of an automated reader as shown in FIG. 4, an operator manually enters the trigger, e.g., manually enters the issue date. Of course, other information may trigger a watermark detection process like an expiration date, issuer code or jurisdictional identifier.

[0098] An authenticator (e.g., a processor under the control of software) uses the trigger to determine whether to carry out a watermark-based authentication procedure. For example, the issue date may correspond with a date that indicates that a digital watermark or a plurality of digital watermarks should be present on an authentic ID document. If so, the authenticator directs a digital watermark verification process, like is discussed in the above embodiments, to determine whether a digital watermark is present. Or the issue date may indicate that the ID document was produced before digital watermarks were rolled out into production (e.g., indicating a so-called legacy card). The authenticator then skips or disables a watermark detection process and/or authenticates the ID document using a different authentication technique.

[0099] In an alternative, but related, implementation, an authenticator directs a watermark detector to analyze optical scan data corresponding to an ID document. The scan data is searched for a digital watermark or for a plurality of digital watermarks. If the watermark detector is unable to find a digital watermark, the authenticator looks for a legacy indicator. For example, an issue date or expiration date is used to determine whether the ID document is a legacy document; or an operator is prompted to enter an issue or expiration date, or to enter a determination as to whether the ID document is a legacy card.

[0100] In another implementation we check for a legacy indicator (e.g., issue date) regardless of whether a watermark is detected. This implementation addresses a counterfeiting scenario where a watermarked feature (e.g., a watermarked photograph) is lifted from an authentic watermarked ID document and pasted onto a legacy document that should not include a digital watermark. If a watermark is found, but the legacy indicator suggests that a watermark should not be present, the ID document is considered suspect.

[0101] A client code can be combined with an issue date or expiration date to indicate which documents should or should not include digital watermarks. A “client code” is broadly used herein and may include, e.g., a code to identify a particular client or a particular type of identification document. The client code can be associated with a particular issue or legacy date, to help identify whether an identification document should or should not include digital watermarking. For example, a client code for a sports arena can be used to signal that all employee badges issued after Jan. 1, 2003 should include a digital watermark or a plurality of digital watermarks. And a client code for an airport tarmac crew can be used to signal that all security badges issued after, e.g., Apr. 15, 2003 should include a digital watermark or a plurality of digital watermarks. A magnetic stripe or barcode (or other machine-readable feature) carries the client code. An authenticator uses the client code to make a determination, for a particular client, as to whether a digital watermark should be present. (An authenticator can manage various client codes and relevant legacy dates via, e.g., a look-up table, software programming or by accessing a local or remote database.). By way of further illustration, an authenticator receives the airport's tarmac crew client code from a security badge. The issue date for the security badge is Mar. 11, 2003. The authenticator then knows that a digital watermark should not be present. Or the authenticator receives a client code corresponding to the sports arena. The corresponding employee badge was issued on Jun. 23, 2003. The authenticator then expects to find digital watermarking on the employee badge. Of course, client codes can be assigned to other entities, like states, nations, companies, etc.

[0102] An authenticator is provided in another implementation to operate primarily between two different modes. Consider FIG. 5, where a versatile authentication process toggles between two authentication branches depending on a watermark detection trigger. The process is carried out, e.g., through a processor or electronic processing circuitry controlled by software. The trigger is analyzed to determine whether an ID document is expected to include digital watermarking embedded therein. If watermarking is expected, optical scan data is analyzed to recover the watermarking. In a two-watermark implementation, payload data is obtained from each of the two watermarks. The payload data, or subsets of the payload data, is compared to determine whether the document is authentic. Of course, if the watermarking is not recoverable, the document is considered suspect (e.g., potentially a counterfeit or unauthorized document).

[0103] If the trigger indicates that the ID document is not expected to include watermarking then the ID document is authenticated using a non-watermarking technique. For example, in one implementation the issue or expiration date (with perhaps a client code) is used as a trigger. If the issue date indicates that watermarking is not expected, another authentication clue, like a bearer's birth date, is obtained from the ID document. A birth date can be machine-read, e.g., from a magnetic stripe, optical memory, electronic circuitry, and/or barcode (e.g., PDF-417, 1-D barcode, etc.). The birth date can also be obtained from a remote repository, e.g., which is interrogated with a document number (e.g., driver's license number). A birth date obtained from such methods (e.g., machine-read, remote access, etc.) is referred to as a “stored birth date.” The stored birth date is then cross-correlated with a birth date that is printed or otherwise carried on the ID document. The printed or otherwise carried birth date is called a “carried birth date.” In some cases the carried birth date is carried via a machine-readable feature. For example, the stored birth date is conveyed through a barcode, while the carried birth date is carried by a magnetic stripe. Retrieving the carried birth date for cross-correlation with the stored birth date can be machine-aided (e.g., with OCR input of a printed birth date), or can be aided by an operator entering the birth date as printed on the ID document. The cross-correlation correlates the stored birth date with the carried birth date. The document is considered authentic when the carried and the stored birth dates correspond. Of course, if either (or both) of the carried or stored birth dates is not recoverable, or is not obtained, the document is considered suspect. Thus, this implementation provides a watermark readable/non-readable authenticator. (Instead of a birth date, other bearer or document specific information like an address (or a hash of an address), name (or hash of a name) or document number (or hash of a document number) can be suitably interchanged. The term “hash” in this context implies a reduced-bit representation of a larger number, value or character string.)

[0104] In still further implementations we add or change a feature on an ID document to signal that the ID document includes a digital watermark. For example, we shift or reposition a photograph carried by the ID document. Instead of positioning the photograph in the upper, left hand corner of the document, we position the photograph in the lower right hand corner, or we offset the photograph by a predetermined spacing. Or we print text (e.g., birth date) in a color (e.g., green) or with a different font. Still further we can place a machine-readable feature in artwork (e.g., a predetermined frequency response due to repetitive spacing of artwork features or lines). These features can be used to trigger (either manually or automatically) watermark detection. Or we can add a predetermined laminate, perhaps embossed or laser engraved features, which can be used to signal watermark detection. Related is a certain type of card stock—perhaps textured with machine-readable code—that signals digital watermarking. Indeed, most of the security features detailed in Appendix A can be used to signal the presence of a digital watermark. In still further implementations, we add a feature that can be pattern recognized. For example, we add a graphic that resembles a triangle. After capturing image data, a pattern recognizer searches the image data for the expected triangle. If found, the triangle signals an expected presence of digital watermarking.

[0105] Sometimes we prefer to add more subtle features to signal the presence of a digital watermark. In one implementation we shift the spatial starting position of text relative to other document features. For example, a document bearer's printed name or employee number can be spatially shifted with respect to the bearer's photograph. The spatial distance between a starting point of text and a photograph (e.g., a lower right hand corner of the photograph) can be machine-measured and used to trigger watermark detection. Of course, instead of shifting text, we can subtly reposition graphics or other artwork as well. In other cases we trigger off of an identification document number. For example, the number may include an extra character (e.g., the seventh or eighth character in) or a leading one or zero can be used to indicate digital watermarking. In other cases we use a plurality of characters, e.g., the last four characters or every other character to indicate the presence of digital watermarking.

[0106] While we have discussed handling legacy documents with respect to digital watermarking, this embodiment of the present invention is not so limited. Of course, our techniques apply to triggering detection of other types of machine-readable features or identifying other types of legacy documents as well.

[0107] A few illustrative combinations are shown below. Of course, these combinations are not meant to limit the scope of this aspect of the invention. Rather, the combinations are provided by way of example only.

[0108] C1. An identification document authenticator operable to authenticate an identification document through a digital watermarking authentication process or through a non-digital watermarking authentication process, a process being selected depending on a detection trigger, said authenticator comprising:

• • [0109] electronic processing circuitry;
  • [0110] a system communications bus;
  • [0111] memory in communication with the electronic processing circuitry via the system communications bus, said memory comprising instructions for processing by the electronic processing circuitry, said instructions comprising instructions to:
    • [0112] analyze data corresponding to the detection trigger;
    • [0113] determine whether the data indicates an expected presence of digital watermarking in the identification document, and if so indicated, authenticate the identification document through the digital watermarking authentication process; and
    • [0114] determine whether the data indicates an expected absence of digital watermarking, and if so indicated, authenticate the identification document according to the non-digital watermarking authentication process.

[0115] C2. The authenticator of C1, wherein the identification document carries the detection trigger.

[0116] C3. The authenticator of any one of C1 or C2, wherein the detection trigger comprises a document issue date

[0117] C4. The authenticator of any one of C1 or C2, wherein the detection trigger comprises a document expiration date.

[0118] C5. The authenticator of C2, wherein the detection trigger is carried by a machine-readable feature.

[0119] C6. The authenticator of C5, wherein the machine-readable feature comprises at least one of a barcode, magnetic stripe, optical memory and electronic circuitry.

[0120] C7. The authenticator of C1, wherein the non-digital watermarking authentication process comprises a cross-correlation of first and second instances of bearer or document specific information carried by the document.

[0121] C8. The authenticator of C7, wherein the first and second instances each comprise data corresponding to a birth date.

[0122] C9. The authenticator of C7, wherein the first and second instances each comprise data corresponding to a name.

[0123] C10. The authenticator of C7, wherein the first and second instances each comprise data corresponding to a document number.

[0124] C11. The authenticator of C7, wherein the first and second instances each comprise data corresponding to an address.

[0125] C12. The authenticator of claim C1, wherein the digital watermarking comprises a first digital watermark including a first payload and a second digital watermark including a second payload.

[0126] C13. The authenticator of C12, wherein the digital watermarking authentication process comprises at least a cross-correlation of some of the first payload with at least some of the second payload.

[0127] C14. The authenticator of C1, wherein the memory comprises electronic memory circuits.

[0128] C15. The authenticator of C14, wherein the electronic processing circuitry comprises a processor.

[0129] C16. The authenticator of C1, wherein the memory comprises at least one of removable memory and fixed memory.

[0130] C17. The authenticator of C1 further comprising an input device.

[0131] C18. The authenticator of C17, wherein the detection trigger is manually entered into the authenticator via the input device.

[0132] C19. The authenticator of C17, wherein the detection trigger is machine-read into the authenticator via the input device.

[0133] C20. The authenticator of C17, wherein the input device comprises at least one of a key pad, mouse, magnetic stripe reader, optical memory reader, optical sensor, barcode reader, touch screen and smart card reader.

[0134] D1. A method to determine whether to authenticate an identification document through a digital watermark authentication process, wherein the identification document comprises a detection trigger, said method comprising the steps of:

• • [0135] receiving digital data corresponding to the detection trigger;
  • [0136] if the detection trigger indicates an expected presence of digital watermarking, analyzing optical scan data to decode the digital watermarking; and
  • [0137] if the detection trigger indicates an expected absence of digital watermarking, providing an indication that the identification document is a legacy document.

[0138] D2. The method of claim D1, wherein the trigger comprises a document issue date.

[0139] D3. The method of claim D1, wherein the trigger comprises a document expiration date.

[0140] D4. The method of any one of D1-D3, wherein the document comprises a magnetic stripe, and wherein the trigger is stored by the magnetic stripe.

[0141] D5. The method of any one of D1-D3, wherein the document comprises electronic circuitry, and wherein the trigger is stored in the electronic circuitry.

[0142] D6. The method of any one of D1-D3, wherein the document comprises a machine-readable feature, and wherein the trigger is conveyed through the machine-readable feature.

[0143] D7. The method of D6, wherein the machine-readable feature comprises a barcode.

[0144] D8. The method of D1, wherein the trigger comprises a client code.

[0145] D9. The method of D8, wherein the trigger further comprises at least one of a document issue date and a document expiration date.

[0146] D10. The method of D1, wherein the providing an indication comprises disabling or foregoing a watermark detection process.

[0147] D11. The method of D1, wherein the providing an indication comprises outputting a signal to indicate that the document is a legacy document.

[0148] D12. The method of D11, wherein the signal is displayed through a graphical user interface.

[0149] D13. The method of D11, wherein the signal activates a warning.

[0150] D14. The method of claim D11, wherein the signal controls or initiates an automated, but non-watermarking based, authentication process.

[0151] Sixth Embodiment—Plural Watermarks

[0152] While some of the above embodiments have envisioned enhanced identification document security through embedding of two digital watermarks, the present invention is not so limited. Indeed, the inventive techniques discussed herein will apply to identification documents including three or more watermarks as well. For example, watermark payload correspondence as discussed under the Second Embodiment can involve three or more watermarks.

Advantages of Watermarking Identification Documents

[0153] Some advantages provided to identification documents by digital watermarking may include the following bullet list. Of course, this list is not meant to limit the present invention, and many of the disclosed embodiments need not include such advantages. By way of illustration only, some watermarking advantages include:

• • [0154] A covert security feature, since a watermark is not generally visually discernible.
  • [0155] No significant impact on an identification document's limited “real estate.”
  • [0156] When deployed across multiple areas, a simple, common authentication mechanism (watermark) is provided when optically reading the identification document. This addresses the complexity issues related to detecting false documents. In a driver's license context, both documents with and from out of state can be authenticated.
  • [0157] Linking to a database (e.g., a DMV, insurance database, etc.) can be enabled, based on policies set and enforced by each document issuing government or organization. Watermark reading software can be implemented to provide access to software based on government or organization policy and reading audience, or alternatively, in-state/out-of-state status.
  • [0158] By providing a linking mechanism versus encoding information on the document itself, inappropriate access to cardholder personal data is prevented. Also, if a government's policy regarding data access for a certain audience changes, this can be implemented in the reader software, enabling dynamic policy changes, without having to reissue identification documents.

[0159] Some advantages of watermarking to government agencies (e.g., DMV, law enforcement officials, and other authorized audiences) and financial institutions are discussed in the following bulleted list. Of course, this list is not meant to limit the present invention, and many of the disclosed embodiments need not include such advantages. By way of illustration only, some watermarking advantages for government agencies include:

• • [0160] DMV forensics agents can detect cards that have been copied and altered, as well as track back to the originating card(s).
  • [0161] Point of sale agents for age-controlled products can locally authenticate a card with respect to card authenticity and age status, without gaining machine-readable access to the individual's demographic data and compromising cardholder's privacy. Watermarking may also provide an ability to read cross-jurisdictional identification documents thus providing better data sharing.
  • [0162] Any inspector equipped with a digital watermark reader can detect “simulated” cards.
  • [0163] The common “photo swap” attack can be automatically detected (e.g., see the first and second embodiments discussed above).
  • [0164] Text alterations and/or substitutions on otherwise legitimate documents can be automatically detected.
  • [0165] Insurance verification can be provided to Law Enforcement by linking to an insurance company database, using the watermark identifier information.
  • [0166] Law enforcement or other authorized users, with the appropriate authorizations and privileges, can selectively and securely link to “back-end” databases for access to photos, biometrics, demographics, and outstanding warrants via a simple imaging or scanning of the identification document.
  • [0167] Crosschecks on a cardholder's name can be done so that financial institutions can further authenticate identity.
    Detecting Media Areas Likely of Hosting Watermarks

[0168] Orientation Components

[0169] The subject matter in this section relates to assignee's U.S. patent application Ser. No. 10/032,282, filed Dec. 20, 2001 (published as US 2002-0114491 A1) and Ser. No. 09/945,244, filed Aug. 31, 2001 (published as 20020057823 A1). Each of these U.S. patent documents is herein incorporated by reference

[0170] Steganographic calibration signals (sometimes termed “marker signals,” reference signals,” “grid signals,” or “orientation components,” etc.) are sometimes included with digital watermarking signals so that subsequent distortion of the object thereby marked (e.g., a digital image file, audio clip, document, etc.) can later be discerned and compensated for. Such arrangements are detailed in the related applications.

[0171] One type of watermark orientation component is an image signal that comprises a set of impulse functions in a transform domain, like a Fourier magnitude domain, e.g., each with pseudorandom phase. To detect rotation and scale of a watermarked image (e.g., after printing and scanning of the watermarked image), a watermark decoder converts the watermarked image to the Fourier magnitude domain and then performs, e.g., a log polar resampling of the Fourier magnitude image. A generalized matched filter correlates a known orientation component with the re-sampled watermarked signal to find the rotation and scale parameters providing the highest correlation. The watermark decoder performs additional correlation operations between the phase information of the known orientation signal and watermarked signal to determine translation parameters, which identify an origin of the watermark signal. Having determined the rotation, scale and translation of the watermark signal, the watermark reader then adjusts the image data to compensate for this distortion, and extracts the watermark message signal.

[0172] An exemplary orientation signal may include one or more of the following characteristics:

• • [0173] 1. It comprises a collection of impulse or delta functions in the Fourier magnitude domain;
  • [0174] 2. The impulse functions have pseudo random phase (i.e. the phase is random, yet the phase is known so that translation (e.g., its X and Y origin) of the watermark can be computed by correlating the know phase information of the calibration signal with the watermarked signal); and
  • [0175] 3. The impulse functions are typically distributed in the mid-frequency range so as to survive distortion yet not be perceptible

[0176] In other embodiments, different orientation components can be used in differently watermarked excerpts of the content. An image may be segmented into blocks, pre-filtered, and then converted into the Fourier domain. The Fourier representation for all the component blocks can be accumulated, filtered, and remapped into the log-polar domain.

[0177] Multiple Watermarks on Printed Documents

[0178] As discussed above, some watermarking implementations include a printed document having a plurality of digital watermarks embedded thereon. We have developed methods and systems to quickly identify areas where the digital watermarks may be embedded, and then focus decoding efforts on the identified areas, particularly when each of the embedded digital watermarks includes an orientation component (or calibration signal). Message decoding efforts can focus on areas identified as likely candidates of including a digital watermark.

[0179] We choose to illustrate our inventive methods and systems with respect to printed documents, such as financial documents, e.g., checks, banknotes, financial instruments, legal instruments and identification documents. Some financial and identification documents may include so-called radio frequency identification circuitry (RFID), smart card chips or circuitry, optical memory, magnetic memory, etc. Of course, our techniques apply to other objects and media (e.g., digital images and video) as well.

[0180] FIG. 6 illustrates a printed ID document 100. ID document 100 preferably includes a photographic representation 102 of a bearer of the ID document 100. The photographic representation 102 may include, e.g., a background portion 102a and a likeness of the bearer portion 102b. The ID document 100 will typically include printed indicia like text (e.g., name, address, jurisdiction, document number, birth date, etc.), machine-readable code (e.g., 1 or 2-D barcodes—not shown in FIG. 6), graphics, background patterns, seals, emblems, artwork, etc. Sometimes the printed indicia is printed with visible or invisible (e.g., UV or IR) inks. Of course, the ID document 100 may include a wide variety of other features like optical or magnetic memory, microprinting, holograms, Kinograms®, electronic circuitry (e.g., a so-called smart card), ghost or faintly reproduced images, etc., etc.

[0181] ID document 100 includes a plurality of watermarks embedded thereon. To simplify the discussion ID document 100 includes two (2) digital watermarks. It should be appreciated, however, that our techniques apply to documents including one and three (3) or more digital watermarks as well.

[0182] A first digital watermark is embedded in the photographic representation 102 (FIG. 6). The first digital watermark can be embedded in the background portion 102a, embedded in the likeness portion 102b, or embedded in both portions 102a and 102b. In our preferred implementation we redundantly embed the first digital watermark throughout the photographic representation 102 or a portion of the photographic representation 102. The first digital watermark typically includes an orientation component (hereafter called a “first orientation component”) and a first message component. The first orientation component is embedded at (or is embedded so as to represent) a first orientation. For example, the first orientation corresponds to a first scale, rotation and/or translation. Translation can be a measure or reflection of a distance or relationship to a predetermined origin or spatial location (e.g., the upper left corner of the identification document). An orientation component may also reflect a particular message starting position or location. If the first digital watermark is redundantly embedded throughout the photographic representation 102, the redundant instances of the first orientation component preferably include the same (or at least closely similar) first orientation.

[0183] A second digital watermark is embedded in area 104. Area 104 includes a graphic, seal, background area, artwork, ghost image, photographic image, hologram, Kineogram® or line art, etc. (hereafter generally referred to as “artwork”), in which the second digital watermark is embedded. The artwork may overlap with text or other document features. While area 104 is illustrated as a bounded and limited area, area 104 can alternatively occupy much larger or smaller document 100 areas. In our preferred implementations we redundantly embed the second digital watermark throughout area 104. The second digital watermark typically includes an orientation component (hereafter called a “second orientation component”) and a message component. The second orientation component differs from the first orientation component. The difference may be subtle such as a difference in embedding scale, rotation or translation, or may explicitly differ in terms of a different calibration technique or embedding protocol.

[0184] The second orientation component is preferably embedded at (or is embedded so as to represent) a second orientation, e.g., at a second scale, rotation and/or translation. If the second digital watermark is redundantly embedded throughout area 104, the redundant instances of the second orientation component preferably include the same second orientation.

[0185] Thus, the first and second digital watermarks each preferably include an orientation component that differs from one another (e.g., the orientation components differ in terms of scale, rotation and/or translation, etc.) or differs relative to a know or predetermined orientation component.

[0186] A digital watermark detection process is discussed with respect to FIGS. 7 and 8A-8C. ID document 100 is presented to a sensor 110 (FIG. 7). Sensor 110 can take a variety of forms. For example, sensor 110 may include a flatbed scanner, an optical senor or array, a CCD optical sensor, a handheld device, web camera, a digital camera (e.g., as is commonly associated with today's cell phones or handhelds), etc. Sensor 110 captures scan data (or “image data”) that represents ID document 100 or a portion of the ID document 100. The captured scan data is communicated to a watermark detector 120 (e.g., electronic processing circuitry controlled by software).

[0187] Watermark detector 120 is configured to analyze windows or image areas (e.g., blocks, image portions or segments), in search of watermark orientation components or in search of characteristics that are associated with watermark orientation components. For example, the watermark detector 120 analyzes scan data associated with a first window 60a (FIG. 8A). The window sizes can vary according to detector specifications. To illustrate, we select a window, e.g., having 32×32 pixels, 64×64 pixels or 128×128 pixels.

[0188] We prefer a sliding window approach for our detector. That is, after the watermark detector 120 analyzes image area 60a, it analyzes an adjacent, perhaps even overlapping, window area 60b, and so on until it analyzes a last window 60n for a first window row (FIG. 8B). The window can be repositioned to analyze additional windows (e.g., 62a-62n) until it analyzes a last row of windows (64a-64n) as illustrated in FIG. 8C. Of course, the detector 120 can alternatively select windows at random (or pseudo-random), or select windows corresponding to a central area of the ID document 100. Also, a first window 60a need not be selected from the upper left hand corner of the ID document 100, but can be selected from other starting image portions as well. In still further alternative implementations, once a watermark has been found in a particular window, the detector analyzes adjacent windows, and then branches out from the adjacent windows.

[0189] While not necessary to do so, the watermark detector 120 preferably analyzes all of the captured image data. (Alternative implementations only look at a subset of the captured image data.). To illustrate, e.g., if the scan data corresponds with a 1024×1280 pixel area, and if 64×64 pixel windows overlap by 50 percent, about 1280 windows are analyzed. Of course the number of windows will decrease as the window size increases and/or as the overlap area decreases.

[0190] Each window or image portion is preferably classified in terms of its orientation characteristics. For example, the detector 120 compares an orientation of each image window or block (some of which may include no watermark signal at all) to that of a predetermined orientation (e.g., an orientation corresponding to an expected watermark orientation component). A metric or measure can be used to classify each image portion in terms of its determined orientation or in terms of its determined orientation relative to the predetermined orientation. An orientation measure may include factors such as rotation, scale, translation, etc.

[0191] The detector 120 (or a computer cooperating with the detector 120) remembers or groups together those windows or image portions which have the same metric or orientation measures. The term “same metric” is broadly used and may include an exact sameness or the metric may include a tolerance. For example, orientations falling within plus-or-minus 1-10 degrees (for rotation) may be grouped together, or those falling within plus-or-minus 2-10% scale or translation might similarly be grouped. After grouping the blocks, the detector (or a cooperating computer) sifts through the groupings to identify predominate groups. The two most predominate groups will likely correspond to one of a first watermark orientation component and a second watermark orientation component.

[0192] FIG. 9A shows a histogram illustrating a relationship between classified blocks (where the z axis represents a number of image windows having a particular orientation) in terms of scale (x axis—in terms of a scale relative to a predetermined scale) and rotation (y axis—in terms of rotation angle in degrees from a predetermined angle). As shown, the histogram includes two strong peaks, each at different orientations. To be more precise, the two peaks share the same rotation, but differ in terms of scale. The peaks correspond to the first watermark orientation component at a first rotation and first scale, and a second watermark orientation component at the first rotation, but second and different scale. The histogram provides a tidy graphical analysis to determine blocks with the same (or closely similar) orientation values. The small peaks in FIG. 9A correspond to image windows that include various orientation values (e.g., images areas that do not have a digital watermark orientation component embedded therein).

[0193] There may be some cases in which the two or more watermark components have the same rotation and scale. For example, only a single predominate peak is seen in FIG. 9B since the first and second orientation components include the same rotation and scale. In these cases other orientation parameters, like translation, can be analyzed to distinguish between different orientation components. These other implementations may include an orientation metric, which includes a translation factor. Image portions can be plotted or analyzed in terms of their orientation metrics as shown in FIGS. 10A-10D. More precisely, FIG. 10A shows an identification document including two watermarks redundantly embedded therein (shown by the two “grids”—one solid and one dashed—where each grid tile represents a redundant instance of a watermark or watermark component). Each of the watermarks includes a different translation with respect to an upper left corner (or other predetermined origin) of the identification document. As a watermark detector analyses the document as discussed with reference to FIGS. 8A-8C it may determine a particular translation value for each window. Since the redundant instances will have the same translation values (perhaps after adjustment for relative tile positioning within the first or second grids), watermarked areas can be determined. FIGS. 10B and 10C illustrate document areas having the same relative translation in terms of x-translation and y-translation. Windows can be grouped together based on similar translation values as shown by the histogram in FIG. 10D. The predominate two groups identify likely locations of the first and second watermarks, since they have common translation values.

[0194] Once predominate orientations are identified, document areas 80 and 82 shown in FIG. 11A (or areas corresponding to scan data including the predominate orientations) can be identified. The identification may involve, e.g., determining which image portions have an orientation corresponding to predominate orientations or which image portions fall within the predominate groups. Areas 80 and 82, or image windows corresponding to areas 80 and 82, can be further analyzed to detect a watermark message. (Areas 80 and 82 in FIG. 11A are intentionally drawn with jagged or varying boundaries. The variation may be caused in some implementations, e.g., where the sample windows do not precisely correspond with embedding areas boundaries.).

[0195] Sometimes we may see an isolated window 84 identified as shown in FIG. 11B. The isolated window may be identified as likely including a watermark because is has an image or noise orientation that is similar to one of the selected watermark orientation components. We optionally test for isolation to jettison such an isolated block 84, since the isolated block is not likely to include watermark information, but rather reflects noise having a similar orientation. One method tests each selected window to ensure that it has at least one or more neighboring selected windows. If it does not, the isolated window 84 is disregarded.

[0196] As an alternative grouping technique, a watermark “strength” is determined for each window. Each image window then has an orientation metric and a strength metric associated therewith. The strength metric is a representation of a watermark characteristic, or a relative correlation between an expected watermark signal and the detected watermark signal. Windows are grouped according to their orientation metrics, but are represented according to a collective strength of all of the windows in a group (e.g., we add together the strength metric for each window within a group). Groups are then represented in terms of collective group strength and not in terms of the number of windows within a group. A strength metric provides an even more prominent indication of watermarked windows, e.g., as shown by the histogram peaks in FIG. 12.

[0197] As a further alternative approach, we filter image data or histogram data to help jettison unwanted noise. This approach even further increases the prominence of window groups or peaks over background noise.

[0198] Our approach saves processing time since only those areas including watermark orientation components are further analyzed to detect a watermark or payload message.

[0199] Copy Detection

[0200] In addition to reducing processing time and ensuring better watermark detection, our techniques may be used as a copy detection tool.

[0201] Consider a financial document like a check. A common counterfeiting technique “cuts and pastes” (perhaps digitally) features from one check to another check. For example, a counterfeiter may cut and paste a bank seal or logo from one check onto another.

[0202] To help detect a copy we provide a document (e.g., a check from Bill's Bank) with two embedded digital watermarks. Each of the first and second digital watermarks includes a first and second orientation component, respectively. The first and second orientation components have a known relationship to one another or a known relationship to a predetermined orientation component. To simplify the discussion, lets say that the first orientation component has a scale that is 82% of an expected scale, and the second orientation component has a scale that is 78% of an expected scale.

[0203] A watermark detector can expects to find the relative scales for the first and second orientation components in order to authenticate Bill's check. (Of course, the detector can account for any optical sensor discrepancies, such as unwanted scaling relative to the printed check.).

[0204] The document is considered suspect if the orientation components' scales are not as expected.

[0205] In some implementations, each bank (or client type) includes a unique orientation component relationship between its watermarks. For example, if the customer or account is associated with a commercial endeavor, then the orientation components are related according to a first predetermined relationship. But if the customer or account is associated with a government endeavor then the orientation components are related according to a second predetermined relationship, and so on.

[0206] A third watermark or watermark component can be used to convey the predetermined relationship or a watermark detector can be configured to expect a certain relationship.

[0207] Even if the counterfeiter is careful when replicating a predetermined scale and rotation, it remains difficult to properly align the multiple watermarks to achieve a predetermined translation.

[0208] Our copy detection also works with ID documents where a counterfeiting technique involves photo or feature swapping. A first watermark is embedded in a photograph, and a second watermark is embedded in a background or artwork. Each of the first and second watermarks includes an orientation component. The orientation components have a predetermined relationship to one another or to a control orientation component. Scan data is collected and orientation parameters are determined. The determined orientation parameters are analyzed to detect a counterfeit document or photo swap. (Different issuers (e.g., state or country) can include a unique orientation component relationship. If the relationship is not what is expected for the issuer, then the document is considered suspect.).

[0209] In alternative implementations a financial document or ID document includes a visible fiducial or overt structure printed or engraved thereon. Orientation parameters are determined relative to the visible fiducial or structure. For example, a scale or rotation of a watermark orientation component is determined relative to the fiducial, and image portions are classified by their relative relationship (or any deviation from an expected relationship) to the fiducial.

[0210] A few illustrative combinations are shown below. Of course, these combinations are not meant to limit the scope of this aspect of the invention. Rather, the combinations are provided by way of example only.

[0211] E1. A method to detect a copy of a printed document, the printed document comprising a first area and a second area including, respectively, a first digital watermark and a second digital watermark, wherein the first digital watermark includes a first orientation component and the second digital watermark includes a second orientation component, said method comprises:

• • [0212] receiving scan data associated with the printed document;
  • [0213] segmenting the scan data into a plurality of portions;
  • [0214] determining an orientation parameter for at least some of the portions;
  • [0215] identifying from the determined orientation parameters, at least two most prominent orientation parameters;
  • [0216] comparing at least one of: i) a relationship between the two most prominent orientation parameters; ii) the two most prominent orientation parameters to predetermined orientation parameters; and iii) a relationship between the two most prominent orientation parameters and an expected relationship between the two most prominent orientation parameters; and based on the comparing step, determining whether the printed document is an original or is a copy of the printed document.

[0217] E2. The method of E1 wherein the printed document comprises a financial document.

[0218] E3. The method of E1 wherein the printed document comprises an identification document, and wherein the first area corresponds with a photographic representation of a bearer of the identification document.

[0219] E4. The method of E1, wherein the orientation parameter comprises rotation and scale.

[0220] F1. A method of determining areas of media that have a likelihood of including digital watermark information, said method comprising:

• • [0221] receiving digital data corresponding to the media;
  • [0222] analyzing a plurality of portions of the digital data to determine an orientation measure for each of the portions, wherein the orientation measure provides an orientation measure relative to a predetermined orientation measure; and
  • [0223] identifying a predetermined number of orientation measures that have at least one of: i) a highest number of portions associated therewith; and ii) a strongest collective watermark strength associated therewith.

[0224] F2. The method of F1, wherein the media comprises a digital image.

[0225] F3. The method of F1, wherein the media comprises video.

[0226] F4. The method of F1, wherein the media comprises a printed document.

[0227] F5. The method of F4, wherein the printed document comprises at least one of an identification document and a financial document.

[0228] F6. The method of F1, further comprising analyzing portions that correspond to the identified orientation measures to recover the digital watermark information

[0229] F7. The method of F6, wherein the identified orientation measures respectively correspond to different digital watermark orientation components.

[0230] F8. The method of claim F6, wherein prior to said analyzing step, said method further comprises manipulating the portions that correspond to the identified orientation measures to compensate for image distortion identified by the orientation measures.

[0231] G1. A method of identifying areas of image data that are likely to include a digital watermark, the image data corresponding to a document comprising a first area and a second area including, respectively, a first digital watermark and a second digital watermark, wherein the first digital watermark includes a first orientation component and the second digital watermark includes a second orientation component, and wherein the document further comprises a visible fiducial, said method comprises:

• • [0232] segmenting the image data into a plurality of portions;
  • [0233] determining an orientation parameter for at least some of the portions relative to the visible fiducial;
  • [0234] identifying from the determined orientation parameters, at least two prominent orientation parameters, wherein portions of the image data that corresponds to the prominent orientation parameters are identified as areas of image data that are likely to include a digital watermark.
    Introduction to Appendix A

[0235] Reference is now made to the accompanying Appendix A, which is herein incorporated by reference. Appendix A details various additional identification document security features that can be used in combination with digital watermarking, including multiple digital watermarks. Indeed, use of these additional security features provides a layer-security approach—making it even more difficult for a potential forger to successfully replicate an identification document.

[0236] In addition to providing a layered security feature, we note that many of these security features can cooperate with a digital watermark. For example, a radio frequency-based security feature may include a code for comparison with (or to unlock or decrypt) a code or message carried by a digital watermark, or vice versa.

[0237] As a further example, a fragile or semi-fragile digital watermark can be embedded or laser engraved in a security laminate. If the laminate is tampered with (e.g., manipulated, removed and/or replaced) the fragile watermark will be destroyed or predictably degraded. In addition, a watermark may be embedded in many of the security features discussed in Appendix A, like fine line printing (background), holograms, optical watermarks, seals and spot colors, to name but a few.

[0238] Still further, one or more digital watermark may be embedded with ultra-violet inks, optically variable inks, specialized inks, infrared inks, etc. In some cases, we anticipate embedding a first digital watermark with conventional ink, while we embed a second digital watermark with one of the specialized (e.g., UV, IR, optically variable, etc.) inks described in Appendix A. The two digital watermarks may cooperate with each other, or may correspond with text (or microprinting) that is providing on an identification document substrate.

[0239] A watermark can also cooperate with biometric information carried by the identification document. In a first implementation, the digital watermark includes a payload having a key to decrypt or decode biometric information stored in a 2-D barcode or magnetic or RF storage carried on the card. In a second implementation, the digital watermark includes information that is redundant with biometric information carried by another security feature. Comparison of the digital watermark information and the biometric information reveals whether the identification document is authentic. In a third implementation, the digital watermark includes at least a portion of a biometric template. Once decoded, the biometric template is used to help authenticate the identification document or to convey information.

[0240] Of course additional combinations of these security features are anticipated. In some cases, the combination will include digital watermarking, and in other combinations they will not include digital watermarking.

[0241] For example, possible combinations might be:

[0242] H1. A security document comprising:

• • [0243] a substrate;
  • [0244] a first graphic carried by the substrate, the first graphic conveying a photographic image to human viewers thereof,
  • [0245] the first graphic being steganographically encoded to convey first plural bits of digital data recoverable by computer analysis of said first graphic; and
  • [0246] a second graphic carried by the substrate, the second graphic conveying a visual image to human viewers thereof, wherein the second graphic is steganographically encoded to convey second plural bits of digital data recoverable by computer analysis of said second graphic;
  • [0247] and a security feature carried by the substrate.

[0248] H2. The security document of H1, wherein the steganographically encoded first plural bits of digital data and the steganographically encoded second plural bits of digital data cooperate to verify authenticity of the security document.

[0249] H3. The security document of H1, wherein the security feature comprises at least one of a deliberate error, a known flaw, fine line background, ghost image, laser encoded optical image, laser engraving, laser perforation, microprinting, a Moiré Pattern, overlapping data, rainbow printing, and security code printing.

[0250] H4. The method of H1, wherein the security feature comprises ultra-violet ink.

[0251] H5. The method of H4, wherein the steganographically encoded first plural bits of digital data and the steganographically encoded second plural bits of digital data cooperate to verify authenticity of the security document, and wherein at least one of the first plural bits of digital data and the second plural bits of digital data cooperate with the security feature to verify the authenticity of the security document.

[0252] The foregoing are just exemplary implementations of the present invention. It will be recognized that there are a great number of variations on these basic themes. The foregoing illustrates but a few applications of the detailed technology. There are many others.

[0253] To provide a comprehensive disclosure without unduly lengthening this specification, applicants incorporate by reference, in their entireties, the disclosures of the above-cited patents and applications. The particular combinations of elements and features in the above-detailed embodiments are exemplary only; the interchanging and substitution of these teachings with other teachings in this application and the incorporated-by-reference patents/applications are also contemplated.

[0254] The section headings in this document are provided for the reader's convenience, and are not intended to impose limitations on the present invention. Features disclosed under one section (or embodiment) heading can be readily combined with features disclosed under another section (or embodiment) heading.

[0255] While some of the preferred implementations have been illustrated with respect to identification documents the present invention is not so limited. Indeed, the inventive methods can be applied to other types of objects as well, including, but not limited to: checks, traveler checks, banknotes, legal documents, printed documents, in-mold designs, printed plastics, product packaging, labels, photographs, etc.

[0256] Also, while some of the implementations discuss embedding first and second digital watermarks, an alternative implementation embeds a single watermark having a first payload component and a second payload component. The first payload component can be embedded, e.g., in a photograph and the second payload component can be embedded, e.g., in a graphic or artwork. The first payload component and the second payload component can be cross-correlated or intertwined to evidence an authentic identification document, as discussed in the above implementations and embodiments.

[0257] Although not belabored herein, artisans will understand that the systems and methods described above can be implemented using a variety of hardware and software systems. Alternatively, dedicated hardware, firmware, or programmable logic circuits, can be employed for such operations. Also, some implementations described in the detailed description suggest that some of the methods or functionality can be carried out using computers or electronic processing circuitry executing software. While this may be a preferred implementation, the present invention is not so limited. Indeed the methods and functionality may be achieved by other means, such as dedicated hardware, firmware, programmable logic circuits, etc.

[0258] In view of the wide variety of embodiments to which the principles and features discussed above can be applied, it should be apparent that the detailed embodiments are illustrative only and should not be taken as limiting the scope of the invention. Rather, we claim as our invention all such modifications as may come within the scope and spirit of the following claims and equivalents thereof.