Method and Apparatus for the Automated Identification of Individuals by the Nail Beds of Their Fingernails

This invention relates to human identification technology, particularly to a method and apparatus for producing identifying data unique to a person through examination and characterization of the person's fingernail bed or other distinguishing anatomical part.

Prior-art personnel identification schemes rely on pictorial recognition, voice printing, fingerprinting, DNA fingerprinting, and retinal scanning, among others. In general there are many means which measure unique physical properties of the person or environmental tracings created by the person. Each means, however, has one or more drawbacks or disadvantages.

Pictorial recognition is easily accomplished by human beings. Recognition of one person by another does not provide a practical solution to the problem of identifying people at large. It can also be accomplished by computers; however computer recognition generally takes longer and is less reliable. A computerized recognition scheme is shown schematically in FIG. 1. A scanning means or camera 10 is used to enter pictorial data 20, usually the subject's face, into computer 30. Recognition algorithms (not shown) are then used to identify the face.

While it has been demonstrated to work in some cases, recognition of faces by a computer is not generally adequate for use in a foolproof identification system. Computer recognition of a face can take some time and requires relatively large amounts of data; hence it may be inappropriate for use in mass screenings. Furthermore, both of these methods can fail if a person changes her or his appearance through the use of makeup, hair coloring, hair styling, accident, surgery, and so forth.

Voice recognition is also easily accomplished by human beings. It can be accomplished by computers; however again it generally takes longer and is less reliable. A computerized voice recognition scheme is shown schematically in FIG. 2. In this case, a microphone 40 is used to enter voice data represented by sound waves 50, and emanating from subject 45, into computer 60. Speech recognition algorithms (not shown) are then used to decipher the speech and match its attributes to one particular speaker.

Voice recognition is presently an inexact science. The algorithms in use today can generally recognize only a relatively few words. This method is unsuitable for mass screening and identification of individuals.

Identification through the use of real or latent (normally invisible fingerprint tracings left on surfaces by body oils, sweat, etc.) fingerprints is now accomplished with ease by both computers and humans.

Real-time fingerprinting of large numbers of individuals is accomplished by automatic scanners in combination with computers. In FIG. 3 a finger 70 is placed in contact with the surface of a scanner 80. Scanner 80 contains a light source (not shown) and photodetector means (not shown) which obtain an image of the subject's fingerprint. Data from scanner 80 are fed to computer 90 for analysis and identification by algorithms (not shown). U.S. Pat. No. 3,864,042 to Leventhal et al, teaches a method of illuminating the surface of a finger and producing an output that is indicative of the fingerprint of an individual. This method requires interpretation and storage of relatively large amounts of data. Also, it carries the stigma of fingerprinting for law enforcement.

Computerized analysis of latent fingerprints is accomplished by dusting the fingerprint tracing with a finely divided powder to make it visible, photographing the print, then scanning the photograph. The scanning process is shown schematically in FIG. 4. A photograph 100 of fingerprint 110 is positioned in front of a camera or scanner 120. Scanner 120 then provides a digitized image of print 110 to a computer 150 for analysis. A fingerprint recognition algorithm (not shown) is used to decipher the familiar loops and whorls contained in the fingerprint. The resulting information can be compared with a database of fingerprints of personnel on file for identification of the person who created the print.

Fingerprints have been used successfully for decades as a means for identifying individuals. Computerized databases have been created and algorithms have been refined so that they can identify an individual through a relatively quick study of the characteristics of the print. Large numbers of fingerprints can be catalogued and accessed quickly. However this means of identification carries the stigma associated with criminals and law enforcement. Furthermore, the analysis involves a two-dimensional study of the individual's print. This study requires and generates moderate amounts of data which must be catalogued and stored.

The term "fingerprinting" in the DNA context is meant to signify a means for identifying individuals through an analysis of their genetic structure. Minute amounts of blood, semen, hair, or body tissues can be used to uniquely identify an individual. The process of DNA fingerprinting is shown diagrammatically in FIG. 5. In this process, a sample 160 of an individual's body tissue or certain fluids is subjected to chemical manipulation, shown schematically as 170. The sample's constituent amino acids are then separated by electrophoresis (separation by an electric field in a viscous fluid) in gel 180. The resulting separation is scanned by scanner 190 and the results analyzed by algorithms (not shown) contained in computer 200. Comparison of electropherograms (traces produced by electrophoresis) can be used to provide positive or negative identification of individuals.

DNA fingerprinting requires intensive laboratory work over a period of days. This method of identification is slow and expensive and so is useful only in certain cases where comparisons are made between only a few individuals. This method is presently not useful for mass screening of individual identities.

Retinal scans involve the scanning or photographing for later scanning of a person's retina. The distribution of blood vessels and various other anatomical structures in an individual's eye is sufficiently complex and unique to permit identification of individuals by comparing the structure of one person's retina with that of another. Retinal scans are accomplished as shown in FIG. 6. Scanner 210 digitizes the image of a retina 220 in an eye 225. The image is sent to a computer 230 for analysis by a retinal recognition algorithm (not shown). However this procedure is too cumbersome and intrusive for mass screening and identification of individuals.

Accordingly, several objects and advantages of the present invention are to provide an improved method and apparatus for rapid identification of individuals. Other objects are to provide a means for identifying individuals which is inexpensive, which does not carry the social stigma of law enforcement, which does not require the storage or manipulation of vast amounts of data, and which is neither cumbersome nor intrusive. Other objects are to provide a means for identification of individuals which is computerized and can yield unambiguous results, and which automatically analyzes fingernail beds.

Additional objects and advantages will become apparent from a consideration of the drawings and ensuing description thereof.

In accordance with the present invention, a method and apparatus are provided which uses a plurality of different light sources to selectively illuminate successive discrete portions of an individual such as fingernails, detect reflected light from this illumination with photodetectors, analyze this detected light, and present results indicative of the identity of the individual.

FIG. 1 is a schematic diagram of a prior-art, computerized human facial recognition system.

FIG. 2 is a schematic diagram of a prior-art, computerized human voice recognition system.

FIG. 3. is a diagram of a prior-art humanly sensible, fingerprint identification system.

FIG. 4 is a schematic diagram of a prior-art, computerized fingerprint recognition system.

FIG. 5 is a block diagram showing some of the steps required in prior-art DNA "fingerprint" analysis.

FIG. 6 is a schematic diagram of a prior-art, computerized retinal scanning system used in personnel recognition.

FIGS. 7A and 7B are drawings of a human fingernail showing a cut-away view of a fingernail nail, the nail bed, and an enlarged section of the structure of the nail bed.

FIG. 8 is a schematic block diagram of a preferred embodiment of the present system.

FIG. 9a is a graphical representation of the unique signature of a subject's nail bed. FIG. 9b shows the amplitude of signals stored at various x'.sub.n locations. FIG. 9c shows a binary representation of the location of loops 360", which are depicted in FIG. 9a.

FIGS. 10A and 10B are front and side views, respectively, of a preferred embodiment of the present system.

FIG. 11A shows the fingernail of a first subject and the data which uniquely identifies this subject. FIG. 11B shows the fingernail of a second subject and the data which uniquely identifies this subject.

FIG. 1--Prior Art

10--Camera

20--Pictorial data

30--Computer

FIG. 2--Prior Art

40--Microphone

45--Subject

50--Sound waves

60--Computer

FIG. 3--Prior Art

70--Finger

80--Scanner

90--Computer

FIG. 4--Prior Art

100--Photograph

110--Fingerprint

120--Camera

150--Computer

FIG. 5--Prior Art

160--Sample

170--Chemical process

180--Electrophoresis gel

190--Scanner

200--Computer

FIG. 6--Prior Art

210--Scanner

220--Retina

225--Eye

230--Computer

FIGS. 7A and 7B--Anatomy of a Fingernail

300--Fingernail

310--Specialized cells

320--Growth plate

330--Cuticle

335--y-axis

337--x-axis

340--Finger

345--Fleshy surface

350--Nail bed

355--Capillary loops

356--Artery

357--Vein

360--Arched portion of nail bed

362--Valley portions of nail bed

FIG. 8--Preferred Embodiment of the Present System

300'--Fingernail

355'--Capillary loops

340'--Finger

362'--Valley portions of nail bed

400--Light source 1

410--Light source 2

420--Thermoelectric cooler

422--Thermoelectric cooler

430--Master oscillator

440--Multiple signal and control lines

450--CCD or photodiode array sensor

460--Analog-to-digital converter

462--Signal lines

464--Signal lines

470--Computer

471--Computer network connection

480--Fiber-optic coupler

490--Lens

500--Acousto-optic, 2-dimensional beam deflector

510--Light beam

520--Sensing ports

530--Input ports

540--Output ports

550--Cartesian coordinate axes associated with finger 340'

560--Cartesian coordinate axes associated with CCD sensor 450

FIGS. 9A, 9B, and 9C--Graphical Analysis of the Data

355"--Capillary loops

360"--Nail bed arches

370"--Nail bed valleys

610--Threshold level

FIGS. 10A and 10B--Preferred Embodiment of the Present System

300'"--Fingernail

340'"--Finger

520'--Sensing ports

700--Chamber

710--Left-hand planar segment

720--Right-hand planar segment

730--Top planar segment

740--Planar segment containing terminuses of sensing ports

FIGS. 11A and 11B--Fingernails of Subjects A and B

300""--Fingernail

340""--Finger

355""--Capillary loops

800--Binary representation of locations of capillary loops, after analysis

Anatomy of the Fingernail and Nail bed--FIGS. 7A and 7B

As shown in FIG. 7A, a human fingernail 300, at the terminus of finger 340 results from the continued regeneration of specialized cells 310 that are pushed out from a growth plate 320 in the direction of y-axis 335. Growth plate 320 lies beneath cuticle 330, at the base of nail 300. Cells 310 comprise primarily the protein keratin, which in nail 300 comprises a hard, flattened, translucent, horny material. As more cells 310 emerge from growth plate 320, fingernail 300 grows and migrates away from cuticle 330 toward the distal end of finger 340. Fingernail 300 grows on top of a nail bed 350. Nail bed 350 rests on a lower, fleshy surface 345. Nail bed 350 is a grooved structure comprising nearly parallel and irregularly separated columns of toughened, blood-perfused flesh extending parallel to y-axis 335.

FIG. 7B shows an enlarged portion of nail bed 350. It has arched portions or lands 360 which contain a large number of capillary loops 355. Loops 355 carry blood between arteries 356 and veins 357, upward and away from a lower, fleshy surface 345 and relatively nearer to the exterior surface of nail 300. Arched portions or lands 360 of nail bed 350 containing loops 355 are separated by grooves or valleys 362, which have varying widths. Valleys 362 lie generally perpendicular to the direction of x-axis 337.

The width, height, and distance between the parallel columns comprising arches 360 and valleys 362 vary with the age and the size of the individual. The width of these columns varies from 0.13 to 0.25 mm (0.005 to 0.01 inch, respectively). The spacings from groove to groove remain proportionally constant over the entire life of the individual, in similar fashion to the loops and whorls of a fingerprint. Thus, unlike facial characteristics, these spacings can be used to identify and distinguish an individual over his or her entire lifetime.

Fingernail Identification System, Preferred Embodiment--FIG. 8.--Overview

The nail bed structures described supra are different for every individual. Therefore the physical and dimensional characteristics of the interface between nail bed 350 and the underside of nail 310 can be employed to uniquely identify individuals. A preferred embodiment of the present system for analyzing and quantifying the nail bed structure is shown schematically in FIG. 8.

In this embodiment, a pair of highly monochromatic light sources 400 and 410 are preferably Light Emitting Diodes (LEDs) or lasers. Sources 400 and 410 may be cooled by thermoelectric coolers 420 and 422. Sources 400 and 410 emit two different and distinct wavelengths of light, preferably .lambda.1=685 nanometers (nm) and .lambda.2=830 nm, respectively. Sources 400 and 410 are both "ON" or active at the same time and are modulated at two different frequencies, typically 100 Hz and 200 Hz, by the respective outputs of a master oscillator 430. This modulation is typically sinusoidal, although other waveforms can be used.

Oscillator 430 also provides synchronized control signals on multiple lines 440 to an acousto-optic, 2-dimensional (2-D) beam deflector 500, a photodiode array-CCD sensor 450, an Analog-to-Digital Converter (ADC) 460, and a computer 470. Computer 470 may be connected to a computer network connection 471. If desirable, computer 470 may communicate with other computers in well-known fashion.

Light from sources 400 and 410 passes through fiber optic coupler 480, lens 490--preferably of the type "graded index" lens--, and acousto-optic beam deflector 500. The resulting, exit light beam 510 is directed into a fingernail 300' by a collection of optical fibers comprising sensing port 520. Only one fiber is illuminated at a time. Although only three optical fibers are shown lying in one row in port 520, it is preferred that there be at least twenty per row and at least twenty rows, comprising an area scanning array.

Deflector 500 preferably delivers the admixed wavelengths, .lambda.1 and .lambda.2 described supra, to one of fibers or input ports 530 at a time. The fiber illuminated by beam 510 is said to be the "illuminated" fiber. Light thus travels from beam deflector 500 through the top, left-hand arm 530 of fiber optic sensing port 520. This light continues downward and exits the illuminated fiber of illuminating and sensing port 520, striking nail 300'. Nail 300' is typically the subject's index fingernail. Light impinging on nail 300' is then absorbed in and reflected from nail 300' and its associated nail bed 350 (FIGS. 7A and 7B) in the manner described infra. Light reflected upward from nail 300' and nail bed 350 (FIGS. 7A and 7B) enters the bottom end of the illuminating sensing port 520. A fraction of the light travels up the right-hand arm or output port 540 of port 520, exiting at the surface of CCD sensor 450. A portion of the reflected light also travels up the left-hand arm 530 of port 520. This portion of light is lost. The fibers in arms or ports 530 and 540 of port 520 are typically 20 to 100 microns in diameter and one to 10 centimeters in length.

A portion of the light from port 520 is reflected from the surface of nail 300'. The remainder of the light passes through nail 300' and is reflected from and scattered within the surface of capillary loops 355 (FIGS. 7A and 7B). Light from source 400, preferably with a wavelength of 685 nm, is largely absorbed by hemoglobin in the blood contained in capillary loops 355'. Loops 355' channel the flow of blood (not shown) between arteries 356 and veins 357 within nail bed 350 (FIGS. 7A and 7B). Light from source 410, preferably with a wavelength of 830 nm, is highly scattered in the presence of the blood cells contained in capillary structures 355 (FIGS. 7A and 7B).

The light of both wavelengths is reflected back through nail 300' and into fibers 520. The ratio of the intensities of this light will depend upon the relative proximity of capillary loops beneath each of the fibers comprising port 520. For example, if the leftmost fiber of port 520 is located directly over one or more capillary loops 355', then the intensity of reflected light at 685 nanometers compared to that of light at 830 nm will have one particular value.

On the other hand, if a central fiber of port 520 is located over a space or valley 362 between loops 355', then the intensity of reflected light at 685 nanometers compared to that of the light at 830 nm will have a lower value. The intermediate case is shown under the rightmost fiber of port 520. It is this difference in light intensity ratios of wavelengths .lambda.1 and .lambda.2 as a function of position which is quantified in the present system.

Light leaving nail 300' enters the illuminated fiber of port 520 and a portion of this light energy is guided into the illuminated member of fibers 540. The distal ends of fibers 540 are optically and mechanically coupled directly to the photodetector elements of area-type Charge-Coupled Device (CCD) sensor 450. One such typical CCD photodetector assembly is the model KAI-1000, manufactured by the Eastman Kodak Company, in Rochester, N.Y., U.S.A. This device contains over one million photodetectors in a plane. Each photodetector is 9 microns square.

In normal CCD operation, light is allowed to fall on each picture element (pixel) of a CCD for a period of time known as the "integration" time or period. This integration time can be compared to the duration of shutter opening in an ordinary, film camera. During this time, photons leaving fibers 540 create charge carriers which are temporarily stored in each pixel region on the CCD, in well-known fashion. At the end of the integration time, all the charge is "read out" from the CCD, one pixel at a time. Signals from CCD 450 are serially presented to the input of ADC 460 via connecting line or lines 362. They are then digitized and a number, representative of the integrated light intensity is presented to the memory of computer 470 via connecting lines 464. One or more pixels on CCD 450 will be illuminated by each of fibers 540. Signals associated with each pixel or group of pixels can be referred to a particular spatial location in the array of fibers 540 and hence in the array of fibers in port 520. The light from each pixel is thus quantified and stored in a particular location in the memory of computer 470. Electronic readout of the signals from CCD sensors is well understood by those familiar with the art of imaging devices and shall not be described in detail here. The Cartesian coordinate axes 560 of the array of fibers in port 520 and the photodetectors on CCD 450 are precisely aligned during manufacture of the assembly.

The integration period of CCD 450 is arranged to be very short compared with the modulation period, typically 0.01 to 0.05 second, of light sources 400 and 410. Moreover, the readout of CCD 450 is coordinated by master oscillator 430 with the modulation of light sources 400 and 410 and the beam positioning by acousto-optic beam deflector 500. Because of this, it is possible to collect two sets of data which can be associated with each of the fibers in port 520. One set is representative of the intensity of light of wavelength 685 nm at each position. The other set is representative of the intensity of light of wavelength 830 nm at each position. After all the individual data at both wavelengths are taken, the above-described intensity ratio can be digitized and stored in the memory (not shown) of computer 470 for each of the fibers in port 520.

Reflected light at wavelengths 680 nm (.lambda.1) and 830 nm (.lambda.2) impinges on CCD sensor 450 at the location of each fiber 540. The integrated intensity value (number of photons) of the light is converted to an electrical signal, typically a voltage at an output connection of sensor 450. In order to conserve output connections, sensor 450 is typically arranged to present the outputs of individual pixels serially, in a row-column fashion. At the end of this integration time, the voltage associated with the number of photons contained in pixel X1, Y1 will be presented at the output of sensor 450, converted to a digital signal by an analog-to-digital converter, and stored in the memory of computer 470. Next, pixel X2, Y1 is digitized and stored. This process continues until all photodetector elements of sensor 450 have been read out, digitized, and stored. This process is completed for both wavelengths .lambda.1 and .lambda.2.

The following data analysis is preferably accomplished by computer 470. This computer is preferably an ordinary personal computer such as the PC 330, sold by International Business Machines (IBM) of Armonk, N.Y., U.S.A. This computer contains 8 megabytes of Random Access Memory (RAM), a floppy disk drive, and a 540 megabyte hard disk drive. This computer may be connected to other computers via a computer network, in well-known fashion. A fingerprint computer is analogous to the preferred computer, although more memory and storage may be required to properly analyze fingerprints. Algorithms written to solve the equations below are well-known to those skilled in the art of scientific computer programming. These algorithms are not discussed in detail here.

In general, each time the individual's finger 340' is scanned, it is likely that it will be oriented at a different angle in the plane formed by axes 550. Therefore the data must first be corrected for angular misalignment between the longitudinal axis (not shown) of finger 340' and the y-axis of the array formed by the fibers in port 520. These fibers are oriented such that the y-axis of the individual's finger is presented in a direction only nearly parallel to the y-axis 560 of CCD 450. Misalignment between the two y-axes is corrected as follows.

The digitized signals at any pixel location are S.sub.Xi,Yj (.lambda.1) and S.sub.Xi,Yj (.lambda.2), where .lambda.1 and .lambda.2 have been defined supra. Xi, Yj denotes any pixel location on CCD sensor 450 which is illuminated by one of the fibers in port 540 associated with the respective fibers in port 520. The normalizing ratio N.sub.Xi,Yj, for i=1 to M and j=1 to M is defined as follows:

Elements N.sub.X1,Y1, N.sub.X1,Y2 . . . N.sub.Xi,YM form a matrix N(x,y) which is a function of discrete pixel coordinates on CCD 450.

Assume that the X-Y axis of CCD 450 (X,Y) is misaligned with respect to the image of the x-y axis (x,y) of nail bed structure 350 (FIGS. 7A and 7B) and that the two axes are oriented at an angle .phi. with respect to each other. It is necessary to rotate reference system coordinates X-Y to new system coordinates x', y' in order to compensate for alignment error. This rotation can be described as follows. Coordinates x' and y' are first resolved into their components in the X, Y plane. Thus

Substituting Eq. 2 into Eq. 1 yields

The values of N(x',y') are quasi-periodic in the x-direction, the direction perpendicular to the longitudinal axis of finger 340. In the y-direction, parallel to the longitudinal axis of finger 340', values of N(x',y') are constant. Thus the alignment error .phi. can be determined from the following derivative equation: ##EQU1## Solving Eq. 4 yields the correct value for .phi.. Substituting the value thus found for .phi. into Eq. 3 yields the corrected value for N(x',y').

The next step in the analysis is the averaging values of N(x',y') along the y' axis for each value of x'. It is not necessary to resolve structural information along y-axis 335 (FIGS. 7A and 7B); only the signal average is required. However it is necessary to properly resolve peak-to-valley distances along x-axis 337 (FIGS. 7A and 7B). Performing this average, while recognizing that the grooves are parallel yields a one-dimensional value for N(x',y'). Hence

The numbers comprising the one-dimensional matrix of output data in Eq. 5 are a mathematical representation of the periodic structure of nail bed 350 (FIG. 7A) and represent the desired, unique identification of an individual.

Graphical Description of Data Analysis--FIG. 9

FIG. 9a shows a perspective view of arches 360" and valleys 362" of nail bed 350 (FIGS. 7A and 7B). The positions of blood-containing capillary loops 355 (FIG. 7B) are shown underneath these structures at x'.sub.1, x'.sub.2, and so forth. Nail 300 (FIG. 7, not shown in FIG. 9) lies above these structures.

FIG. 9b shows the corresponding amplitude of the ratio N(x',y') of signals digitized and stored at each x'.sub.n location, and averaged along the y' direction, as described supra.

If a threshold level 610 is established in the data reduction algorithm employed by computer 470 (FIG. 8), then signals which have amplitude greater than this level can be said to be a logical "1" or Boolean TRUE. Signals which have amplitude less than this level can be said to be a logical "0" or Boolean FALSE. This permits a binary representation of the location of the loops 360 (FIG. 7) and their relative locations x'.sub.n, as shown in FIG. 9c. This representation is the end point of the data reduction wherein a binary data set uniquely identifies an individual by the presence and spacings of loops 355 (FIG. 7B) in the individual's nail bed.

Operation of the Preferred Embodiment--FIGS. 10A, 10B, 11A and 11B.

One arrangement of the preferred embodiment is shown in FIGS. 10A and 10B. In use, the person inserts a finger 340'", preferably an index finger into a chamber 700, as shown. Chamber 700 comprises planar segments 710, 720, 730, and 740 constrains the individual's finger 340'" to lie one orientation. Sensing ports 520' are perpendicular to and terminate in plane 740. In this embodiment, finger 340'" is "upside-down", with fingernail 300'" pressing comfortably against planes 730 and 740; which are preferably oriented at an angle of 45 degrees to force the nail of the inserted finger against segment 740. A gap between segments 730 and 740 permits insertion of a finger 340'" with a long fingernail 300'". Chamber 700 and the array of sensing fibers in port 520' have width equal to or greater than the largest finger which is to be scanned. For simplicity, only twenty-five fibers comprising port 520' are shown in FIG. 10A. As mentioned supra, the preferred embodiment contains many more. After finger 340'" is properly inserted into chamber 700, a sensor (not shown) or an operator (not shown) initiates a scan of fingernail 300 and nailbed 350 (FIGS. 7A and 7B), in the manner described supra. Master oscillator 430 (FIG. 8) turns light sources 400 and 410 "ON". Beam deflector 500 causes the combined light from sources 400 and 410 to illuminate the first fiber in multiple-fiber sensing port 520. This combined light illuminates one discrete portion of nail 300"". Sensing port 520 also receives reflected light from the same location in nailbed 340 (FIGS. 7A and 7B) and transmits this light to Sensor array 450. Sensor array 450 detects light from the first-illuminated fiber and stores electrical charge at that fiber's location. The amount of charge stored is proportional to the intensity and duration of the detected light. Deflector 500 now causes the combined light from sources 400 and 410 to illuminate the second fiber in port 520. Light from the second fiber illuminates a second, discrete portion of nail 300"". Light reflected from this location is stored at a second location in sensor array 450, and so forth until all fibers in fiber port array 520 have been illuminated and their respective data stored in sensor 450. After all fibers have been scanned, master oscillator 430 causes analog-to-digital converter 460 to digitize the data from each position in sensor array 450. These digitized data are stored in computer 470 for subsequent analysis.

Two different fingernails are shown in FIGS. 11A and 11B. One is the index fingernail of Subject A (FIG. 11A) and the other is the index fingernail of Subject B (FIG. 11B). The locations of capillary loops 355'" are shown as lines.

As described supra, columnar data from sensor 340 (FIG. 8) are averaged to provide a single row of data. The positions of fibers in port 520" are shown as rows 800. After thresholding of the data, each position along row 800 contains either a "1" or a "0", or a Boolean TRUE or FALSE. These binary values indicate the presence or absence of a loop 355'" at that position. Thus the identity of Subject A comprises the number string 3, 4, 8, 13, and 14 since these are the indicated positions of loops 355'". It can be seen that the identity of Subject B comprises the number string 3, 4, 5, 7, 12, 13, 16, 17, and 18.

Each scan reduces the data thus obtained from each person's fingernail to an unique number string. This string contains one or more numbers representative of the averaged positions of the capillary loops in that person's fingernail. These number strings are stored in the memory of computer 470 (FIG. 8). Each time a subject's fingernail is scanned, computer 470 can do any of several tasks with the data: it can disregard the data, store the data, or compare the data with all other previously stored number strings, using well-known comparison techniques.

It is thus seen that the present system provides a way to unambiguously identify individuals, preferably by optically scanning their fingernail beds at a plurality of known positions, the illumination being provided at two discrete wavelengths.

Although the preferred embodiment has been described for use with fingernails, other body structures can be used with equal effectiveness. Toenails and locations on the individual's skin can provide useful data similar to that obtained from fingernails. Light of other wavelengths can be used, as well as light from more than two sources. Instead of an acousto-optical scanner, a galvanometer-type scanner can be used. Instead of an area CCD sensor, another type of sensor, such as a television camera electron tube, a photodiode array, or a mechanically scanned linear CCD sensor can be used. In certain cases such as the television camera tube, a lens may be required to image the ends of the sensor fibers onto the photodetector plane. More or less data representative of a person's nail bed can be saved. Instead of a binary condition, the data can be saved as multi-bit, digital bytes or as analog information. Instead of using the master oscillator to control the system, the computer can control the operation of the entire system. Rather than using the absolute row numbers to comprise the identifying number string, a differencing method can be used. In this method, the position of the first capillary loop is regarded as the datum for subsequent "TRUE" row locations. Instead of absolute identification of an individual, a partial match between a stored identity and a new identity can be statistically analyzed. This analysis provides an identity which is qualified as to a statistical level of confidence.

While the present system employs elements which are well known to those skilled in the separate arts of optics, signal analysis and electronics, it combines elements from these fields in a novel way which produces a new result not heretofore discovered.

Accordingly the scope of this invention should be determined, not by the embodiments illustrated, but by the appended claims and their legal equivalents.