Method and Apparatus for Automatically Identifying and Selectively Altering Segments of a Television Broadcast Signal in Real-Time

1. Field of the invention

The invention relates to television signal processing systems and in particular to television signal processing systems for detecting selected segments of a television signal such as commercial advertisement segments.

2. Description of Related Art

It has long been a desire of many television viewers to have a system which automatically detects selected segments of a television signal such as commercial advertisements or other unwanted broadcast segments in real-time during a television broadcast for the purpose of, for example, muting the audio portion of the television broadcast during the unwanted broadcast material. Although a wide variety of techniques have been developed for detecting selected segments of television signals, heretofore there have been no sufficiently effective and economical systems that automatically operate in real-time for muting or similar purposes.

Examples of broadcast segment detection systems include systems for detecting commercial advertisements during the playback of a television program on a video cassette recorder (VCR) for the purpose of controlling the VCR to skip or scan over the commercial advertisements during playback such that a viewer may watch the recorded television program without significant commercial advertisement interruptions. One particularly effective system is described in U.S. Pat. No. 5,333,091 entitled "Method and Apparatus for Controlling a Videotape Player to Automatically Scan Past Recorded Commercial Messages" and sold under the trademark "Commercial Advance." With the system of U.S. Pat. No. 5,333,091, an automatic editing device is coupled between a conventional VCR and a conventional television set. As a broadcast program is recorded on a videotape by the VCR, the editing device detects and records the timing and duration of selected events, such as black frames combined with low audio, that separate segments of the broadcast and creates a time-based map of the selected events. After the recording is completed, the editing device analyzes the time-based map and determines therefrom which segments are commercial advertisement segments and which are program segments. The editing device then records control signals on the videotape prior to and following the commercial advertisement segments. Upon playback, the control signals control the VCR to fast-scan over the commercial advertisement segments so that the recorded broadcast can be viewed substantially without interruption. Related techniques are described in U.S. Pat. No. 5,455,630 and in PCT application PCT/US94/00223 published as WO 94/16442.

Although the systems of U.S. Pat. Nos. 5,333,091 and 5,455,630 and PCT application PCT/US94/00223 are very effective in skipping over commercial advertisements and other unwanted broadcast material during the playback of a recorded broadcast program on a videotape, the systems do not operate in real-time during a received broadcast.

Other examples of broadcast segment detection systems include systems for detecting commercial advertisements during the recording of a television program on a VCR for the purposes of controlling the VCR to skip the commercial advertisements during recording such that a viewer may later watch the recorded television program without any commercial advertisement interruptions whatsoever. An example is a system described in U.S. Pat. No. 4,602,297 entitled "System for Editing Commercial Messages from Recorded Television Broadcasts" which stores and delays the output of a received broadcast signal to a VCR until it can determine whether the delayed portion of the broadcast signal includes commercial advertisements. To this end, the system detects commercial advertisements or groups of commercial advertisements within the delayed portion by detecting amplitude drops in the broadcast signal and determining whether time intervals between the amplitude drops correspond to normal durations of commercial advertisements or groups of commercial advertisements. The system then outputs only those portions of the delayed signal to the VCR that do not correspond to commercial advertisements of groups of commercial advertisements. Hence, the broadcast signal is recorded without commercial advertisements for future playback.

Other examples of systems for detecting commercial advertisements during the recording of a television program are described in U.S. Pat. No. 4,750,052 entitled "Apparatus and Method for Automatically Deleting Selected Program Intervals from Recorded Television Broadcasts" and U.S. Pat. No. 4,782,401 entitled "Editing Method and Apparatus for Commercials During Video Recording", both of which also operate to detect commercial advertisements based upon intervals occurring between amplitude drops or other fades in a received broadcast signal. Rather than storing and deleting the received signals prior to output to a VCR, however, the systems of U.S. Pat. Nos. 4,750,052 and 4,782,401 record the signals on a video tape using a VCR and control the VCR to rewind at the end of each commercial advertisement or group of advertisements such that further program signals are recorded over the commercial advertisements.

None of the aforementioned systems for detecting broadcast segments during the recording of a television program operate in real-time for the purpose of, for example, muting the audio during commercial advertisements.

Still other examples of broadcast segment detection systems include systems for detecting commercial advertisements in broadcast television signals for the purposes of verifying that selected commercial advertisements are being broadcast on the proper days and at the proper times. Examples include U.S. Pat. No. 5,504,518 entitled "Method and System for Recognition of Broadcast Segments" wherein stored digitized segments of broadcast signals are processed by an expert system to determine whether any particular segment corresponds to a selected segment of interest, such as a selected commercial advertisement. Segments that cannot be identified by the expert system, such as new or otherwise unknown segments, are routed to a human operator for verifying that the segment represents a commercial advertisement rather than a portion of a television program. Selected segment information such as the date and time of broadcast of the segments of interest are recorded and tabulated in a report for use by, for example, broadcast advertisers for verifying that the selected segments were broadcast on the proper dates and at the proper times. As part of its analysis of the segments, the expert system compares broadcast signatures with a database of signatures. A variety of techniques are described for defining and detecting signatures. The expert system also considers various cues including the lengths and relative groupings of segments and the intervals therebetween. No real-time analysis appears to be performed by the expert system or by the human operator.

Another system for commercial advertisement broadcast verification is described in Japanese Laid-Open Patent Application No. 1-174192, of Sony Corporation, entitled "Television Signal Processing Unit" wherein an entire days worth of broadcast programs are recorded on videotape. The recorded broadcasts are analyzed to identify commercial advertisement segments by intervals therebetween as detected by luminance signal considerations. The commercial advertisements, and only the commercial advertisements, are then output and recorded on a second videotape. The second videotape thereby contains a days worth of commercial advertisements in extracted form that may be viewed by a human operator for broadcast verification purposes. Again, no real-time detection of commercial advertisements is performed.

Examples of systems for detecting broadcast segments, such as commercial advertisement segments, which have some real-time detection capability include U.S. Pat. Nos. 4,420,769 entitled "Device for the Automatic Editing, at the Receiver, of Unwanted Program Material from Broadcast Electrical Signals" and 4,750,213 entitled "Method and System for Editing Unwanted Program Material from Broadcast Signals". U.S. Pat. No. 4,420,769 describes a system wherein a human operator identifies program segments to be eliminated from a broadcast signal causing a digital code word or signature for the segment to be stored in memory. Subsequently, if the same segment is encountered, the system automatically compares digitized portions of the received signal with the stored signatures to recognize the segment and then to blank out or otherwise eliminate the segment in real-time. One significant problem with this system is that a human operator must first detect program segments to be eliminated. Also, a significant amount of data processing is required to establish and compare the signatures in a reliable manner. U.S. Pat. No. 4,750,213, by the same inventor, sets forth improvements intended, in part, to reduce the processing time required. The system of U.S. Pat. No. 4,750,213 also includes a delay or other storage device for allowing a received signal to be delayed until a determination can be made by the human operator as to whether a particular segment of the received signal should be eliminated. If so, the received, delayed signal is output with the selected segment blanked out or otherwise eliminated. Although U.S. Pat. No. 4,750,213 may represent an improvement over U.S. Pat. No. 4,420,769, it shares the disadvantage that a human operator must detect program segments to be eliminated.

Another example of a system which has some real-time detection capability is U.S. Pat. No. 5,151,788 entitled "Method and Apparatus for Identifying and Eliminating Specific Material from Video Signals" wherein commercial advertisement broadcast segments are distinguished from program segments by the relative level of "activity" within the beginning of each segment. If the activity level is high, the segment is presumed to be a commercial advertisement. The system detects the amount of activity within the first few seconds of a segment following a blank frame and operates to mute or otherwise eliminate the remaining portions of the segment if the detected activity level is representative of a commercial advertisement. The system may control a VCR to stop recording for the remainder of the segment. The patent describes that the optimum time period during which the activity level should be detected is 8 seconds. One disadvantage of the system is that the detection of commercial advertisements based upon an activity level is likely to be unreliable for many broadcasts and, even when the system does correctly detect a commercial advertisement, that detection does not occur until 8 seconds into the commercial advertisement.

Other references that relate to systems for detecting selected segments of a broadcast signal such as commercial advertisement segments of a television signal are described in the following references: U.S. Pat. Nos. 3,919,479 entitled "Broadcast Signal Identification System"; 4,229,765 entitled "Remote Audio and Brightness Control System for a Television"; 4,230,990 entitled "Broadcast Program Identification Method and System"; 4,259,689 entitled "Television Advertising Editing System"; 4,333,110 entitled "Television Editing System"; 4,390,904 entitled "Automatic Circuit and Method for Editing Commercial Messages from Television Signals"; 4,667,466 entitled "Broadcast Program Identification Method and Apparatus"; 4,697,209 entitled "Methods and Apparatus for Automatically Identifying Programs Viewed or Recorded"; 4,739,398 entitled "Method, Apparatus and System for Recognizing Broadcast Segments"; 4,752,834 entitled "Reciprocating Recording Method and Apparatus for Controlling a Video Recorder so as to Edit Commercial Messages from a Recorded Television Signal"; 5,019,899 entitled "Electronic Data Encoding and Recognition System" and European Patent Application Publication No. 0161512 entitled "Program Identification System".

Hence various techniques have been developed for detecting selected broadcast segments such as commercial advertisements within a television signal. None of the techniques described above, however, are effective for reliably and automatically detecting broadcast segments in real-time without the need of a human operator. Accordingly, it would be desirable to provide a simple, inexpensive system which reliably detects selected segments of a broadcast signal without requiring a human operator and wherein, either immediately or after some period of time, all or most selected segments can be detected in real-time.

In accordance with the invention, a method and apparatus, intended to be marketed under the mark Sound!Off.TM., is provided for identifying selected segments of a broadcast signal in real-time for the purpose of, for example, muting the video and audio portions of the signal during the selected segment. Hereinafter "selected segment" refers to any selected portion of a broadcast signal including, for example, unwanted broadcast segments such as commercial advertisement segments. In one embodiment wherein the broadcast signal is a television signal, the method includes the steps of detecting a signature pattern associated with the beginning of a received segment of the television signal and comparing the signature pattern to stored signature patterns representative of selected segments. If the signature pattern matches one of the stored signature patterns, the received segment is thereby immediately identified as being one of the selected segments and can be processed accordingly in real-time to, for example, mute the audio and video portions of the television signal during the received segment as the television signal is output to, for example, a television monitor. If the signature pattern of the received segment does not match any of the stored signature patterns, the received segment is analyzed to determine whether it is nevertheless one of the selected segments and, if so, its signature pattern is stored along with the stored signature patterns. The analysis to determine whether the received segment is one of the selected segments even though its signature does not match any of the stored signatures may be performed by, for example, detecting the length of the received segment upon its completion and then determining whether the length of the received segment matches any of a pre-determined set of permissible lengths for selected segments. If so, the segment is identified as being a selected segment and its signature pattern is stored along with the other stored signature patterns such that, the next time the same segment is received, its signature will then match one of the stored signatures and therefore the received segment will be identified promptly as a selected segment and muted or otherwise altered in real-time. For an embodiment wherein the selected segments are commercial advertisement segments, the permissible lengths are, for example, 15 seconds, 30 seconds, and 60 seconds. The method may be performed by a television set or other device receiving a television signal such as a personal computer television (PCTV).

In one implementation, no stored signatures are initially provided. Hence, the first time any particular selected segment is encountered, it is not immediately identified as being a selected segment and therefore is not muted in real-time. However, the next time the same segment is encountered, the segment will be subject to immediate real-time muting because its signature will have been stored the first time it was received. Thereafter, as long as its signature remains stored, the segment will be subject to real-time muting whenever it is received. As more and more selected segments are encountered and their respective signature patterns stored, the more and more likely it becomes that any particular selected segment encountered while watching a television program will be promptly detected based upon its signature and muted in real-time. Hence, even though no stored signatures are initially provided, a database of signatures is soon developed. Also, in the exemplary implementation, the selected segment detection method may be performed by a television set even while the television set is otherwise not in use to thereby detect and record segment signatures in a background mode to quickly develop a database of signatures. Alternatively, a database of the signatures of common segments, such as common commercial advertisement segments, may be initially provided at the time of manufacture such that the common segments are thereby subject to immediate real-time muting even the first time received.

For many television broadcast systems, commercial advertisements appear in groups of two or more. When used in connection with such systems, the method of the invention preferably operates to not only mute or otherwise alter any detected commercial advertisements of a commercial advertisement group but also to mute or otherwise alter other unwanted broadcast segments appearing within the group, such as station breaks, news bulletins, public service announcements, etc. To this end, the method operates to identify the first of a group of commercial advertisements, then mutes or otherwise alters all segments during a predetermined minimum commercial advertisement group time period. The minimum group time period may be, for example, two minutes and twenty seconds. Hence, all segments appearing during the period are muted in real-time, including commercial advertisements for which the corresponding signature has not yet been stored (and which therefore would not otherwise be subject to immediate real-time muting) and segments that do not match the predetermined permissible lengths (and which therefore would not otherwise be subject to any muting whatsoever), such as the aforementioned station breaks, news bulletins, and public service announcements. Thus the method is not limited to muting only those segments that have matching stored signatures. Rather, the method operates to mute or otherwise alter all segments appearing in a detected commercial advertisement group. However, only the signatures of segments having lengths matching the permissible selected segment lengths are stored. The signatures of a non-commercial advertisement segment, such as a station break, news bulletin, or public service announcement, are stored only if the length of the segment happens to match one of the predetermined permissible segment lengths.

Depending upon the implementation, the method identifies the beginning of a received segment by detecting pre-determined event markers which separate segments. When implemented for use in North America, the method may, for example, identify, a period of one or more black frames combined with low audio as being an event marker representative of the beginning of a segment. When implemented for use in Europe, the method may further identify periods of one or more colored frames, such as blue frames, combined with low audio as being an event marker. When implemented for use in Japan, the method may instead identify a frame break combined with a period of low audio as being an event marker. In any case, once an event marker denoting the beginning of a broadcast segment is detected, the method promptly detects a signature from the segment for comparison against stored signatures. The signature may, for example, be a binary hash code representative of whether selected lines of a selected frame near the beginning of the segment exceed pre-determined luminance threshold levels. By detecting the signature near the beginning of the segment, the signature can be compared against stored signatures to allow the segment to be muted or otherwise altered in real-time before much of the received segment has been displayed. Preferably, the signature is based on frames appearing during the first one second of the segment.

The features, objects, and advantages of the present invention will become more apparent from the detailed description set forth below when taken in conjunction with the drawings in which like reference numerals identify correspondingly throughout and wherein:

FIG. 1 is a block diagram illustrating a television set incorporating a commercial advertisement detecting and muting system configured in accordance with an exemplary embodiment of the invention for detecting a commercial advertisement or other selected broadcast segment in real-time during a television broadcast and for muting video and audio portions of the television signal during the commercial advertisement;

FIG. 2 is a timing diagram illustrating exemplary groupings of consecutive commercial advertisement segments within a television broadcast signal analyzed by the system of FIG. 1 and illustrating event markers separating the broadcast segments;

FIG. 3 is a timing diagram illustrating the muting of luminance and audio portions of a television signal provided by the system of FIG. 1 for a pair of broadcast segments both identified as being commercial advertisements;

FIG. 4 is a timing diagram illustrating exemplary predetermined points in time following a first event marker wherein the system of FIG. 1 searches for a second event marker for the purpose of determining whether the first event marker corresponds to the beginning of a commercial advertisement segment and particularly illustrating a example wherein a second event marker is found at one of the predetermined points in time thereby indicating that the first event marker corresponds to the beginning of a commercial advertisement segment;

FIG. 5 is a timing diagram similar to that of FIG. 4, but wherein a second event marker is not found until after all of the predetermined points in time thereby indicating that the first event marker does not correspond to the beginning of a commercial advertisement segment;

FIG. 6 is a timing diagram similar to that of FIG. 4, but wherein a second event marker is found but not at one of the predetermined points in time thereby also indicating that the first event marker does not correspond to the beginning of a commercial advertisement segment;

FIG. 7 is a timing diagram similar to that of FIG. 4, wherein a second event marker is found at one of the predetermined points in time thereby also indicating that the first event marker corresponds to the beginning of a commercial advertisement segment but wherein yet another event marker, which must be ultimately ignored, is found between the others;

FIG. 8 is a flowchart illustrating an exemplary method performed by the system of FIG. 1 while the television set is in operation for detecting and muting commercial advertisements during a television broadcast wherein, among other features, a signature corresponding to an event marker is compared against a database of signatures corresponding to known commercial advertisements to determine whether the event marker corresponds to the beginning of a commercial advertisement segment;

FIG. 9 is a flowchart illustrating steps performed in furtherance of the method of FIG. 8 to read a segment signature corresponding to an event marker;

FIG. 10 is a representation of a single exemplary test frame of a television broadcast signal analyzed during execution of the steps of FIG. 9 and in particular illustrating exemplary signature test lines of the frame from which the signature is determined;

FIG. 11 is a representation of an exemplary bit string binary hash code signature corresponding to the frame of FIG. 10 and in particular showing exemplary bit values each representative of whether the average luminance of a corresponding test line of the frame exceeds a pre-determined threshold;

FIG. 12 is a flowchart illustrating steps performed in furtherance of the method of FIG. 8 to determine if an event marker nevertheless corresponds to the beginning of a commercial advertisement segment even though the signature corresponding to the event marker does not match any of the signatures of the database;

FIG. 13 is a flowchart illustrating steps performed in furtherance of the method of FIG. 8 to store a signature in the database after it has been determined that the signature corresponds to a commercial advertisement segment;

FIG. 14 is a representation of a signature database accessed during execution of the steps of FIG. 13 wherein the database is configured as a RAM storing a single record of information, including broadcast date/time information, for each unique commercial advertisement signature to thereby allow records to be selectively overwritten as needed based, for example, upon the broadcast date/time information;

FIG. 15 is a diagram similar to that of FIG. 14 but wherein the database is configured as a first-in first-out (FIFO) queue;

FIG. 16 is a diagram similar to that of FIG. 14 but wherein the database is configured to store the signature and broadcast date/time information for each detected broadcast of each commercial advertisement to thereby allow verification that selected commercial advertisements are being broadcast on the proper days and at the proper times;

FIG. 17 is a flowchart illustrating a method for operating the system of FIG. 1 wherein the system is operated even while the television set is otherwise not in use to thereby detect and record commercial advertisement segment signatures in a background mode to thereby quickly develop a full database of signatures;

FIG. 18 illustrates a system wherein an apparatus performing the method described with reference to FIGS. 1-17 is provided external to a television set; and

FIG. 19 illustrates a PCTV configured with hardware or software for performing the method described with reference to FIGS. 1-17.

Referring to the Figures, exemplary embodiments of the invention will be described with reference to a system primarily directed toward detecting commercial advertisements. Initially, an overview is provided with reference to the block diagrams and timing diagrams of FIGS. 1-7. Then, further details of particular implementations are provided with reference to the flowcharts and block diagrams of FIGS. 8-19. The exemplary embodiments are primarily for use in detecting and muting commercial advertisements in a North American television broadcast signal wherein the advertisements are of fixed predetermined lengths and appear in groups. As will be described, the system can additionally mute other unwanted program material segments including station breaks, news bulletins etc. that appear in a detected commercial advertisement group. Moreover, the invention may be alternatively implemented for use with other television broadcast systems, such as those of Europe and Japan, and may be alternatively implemented to identify other types of selected segments of a signal besides commercial advertisements.

FIG. 1 illustrates a television set 100 configured with Sound!Off.TM. and having a receiver 102 and a monitor 104. Television receiver 102, which includes a television tuner, receives an input broadcast television signal from an antenna 106 or from any other appropriate input source such as a coaxial cable input (not separately shown) and outputs a received television signal for eventual display by monitor 104. In this exemplary implementation, the received television signal is an analog NTSC signal including audio, luminance and chrominance components. In other implementations, the television signal is a PAL or SECAM signal or digital signal such as an HDTV signal. Monitor 104 displays the video portions of the television signal and plays the audio portions through speakers (not separately shown).

Rather than routing the received television signal directly to the monitor for display, the television signal is first routed through a commercial advertisement detecting and muting system 108 which operates in real-time to detect target or selected segments of the television signal and to mute selected components of the television signal during those segments prior to output to monitor 104. In the exemplary implementation, the selected segments are commercial advertisements. The muted components of the television signal are the audio and luminance components. As noted, in other implementations, other selected segments may be detected or other components of the signal may be muted or otherwise processed. Commercial advertisement detecting and muting system 108 is, in the exemplary implementation, configured as a single plug-in module having both analog and digital electronics for plugging directly into an otherwise conventional television set. In other implementations, described below, the commercial advertisement detecting and muting system is an after-market device configured for use with a television set having audio and video outputs or is hardware or software provided within a PCTV.

Detection of commercial advertisement segments is achieved as follows. Within detecting and muting system 108, the television signal is first routed through a broadcast segment event marker and signature detector 110 which detects event markers separating broadcast segments of the television signal and, for each segment, detects a signature. A broadcast segment may be a commercial advertisement segment or a portion of a television program, movie, news show, "infomercial" etc. hereinafter referred to as program segments. As noted, the broadcast segments are separated by event markers. In the exemplary implementation which is for use in North America, an event marker is a period of one or more black frames combined with low audio. When implemented for use in Europe, the event marker may also be a period of one or more colored frames, such as blue frames, combined with low audio. When implemented for use in Japan, an event marker is a frame break combined with a period of low audio extending from slightly before to slightly after the frame break.

In any case, commercial advertisement segments may be distinguished from program segments in that commercial advertisement segments almost always appear in groups of two or more and are each of one or more relatively short pre-determined lengths such as 15 seconds, 30 seconds and 60 seconds. The groups are typically at least two minutes long. FIG. 2 illustrates an exemplary television signal having program segments 112, 114, and 116 followed by groups of commercial advertisement segments 118, 120 and 122, respectively, with each segment separated by an event marker 124. Commercial advertisement group 118 includes two commercial advertisement segments of 60 seconds each. Group 120 includes two commercial advertisement segments of 30 seconds and one of 60 seconds. Group 122 also includes two 60 second commercial advertisement segments. Although not shown in FIG. 2, commercial advertisement groups may additionally include other segments such as station breaks, news bulletins, public service announcements etc., which are typically not of the same lengths as commercial advertisements, i.e. the other segments are not 15, 30 or 60 seconds long. If other segments are included, the group is typically at least two minutes and twenty seconds long. The manner by which the non-commercial advertisement segments within a group are processed will be described below.

Each unique broadcast segment has a unique signature. The signature may be any unique characteristic of the segment that can be used to distinguish one segment from another. As will be described below, in the exemplary implementation, the signature is a bit string hash code representative of whether each of a set of selected lines of a selected frame of a received segment has an average luminance level that exceeds a predetermined threshold. The selected frame may be, for example, the tenth frame following the completion of the event marker marking the beginning of the segment. The selected lines may be, for example, the first 64 odd lines of the frame beginning at line 23. Other signatures may alternatively be employed.

Referring again to FIG. 1, as each broadcast segment is received, the signature for that broadcast segment is detected in real-time by detector 110. The signature is routed to a control unit 126 which accesses a signature memory unit 128 to determine whether the detected signature matches any of a set of signatures stored therein known to correspond to commercial advertisement segments. If the detected signature matches one of the stored signatures, then the control unit identifies the broadcast segment from which the signature was taken as being a commercial advertisement segment and forwards a mute control signal to a video/audio mute unit 130. The mute unit also receives the television signal directly from detector 110. In response to the mute control signal, the mute unit mutes the audio and luminance components of the television signal during the commercial advertisement segment and forwards the muted signal to monitor 104. The mute control signal is maintained during the entire commercial advertisement segment. In the exemplary implementation, the signature memory unit stores, for each commercial advertisement signature, the length or duration of the segment. The control unit reads this information and thereby knows when to terminate muting. In other implementations, muting is terminated upon detection of a next event marker. The former technique is preferred because event markers may coincidentally appear in the middle of a commercial advertisement segment. If muting is always terminated upon detection of a next event marker, muting could erroneously be removed in the middle of a commercial advertisement.

Hence, the system mutes the output signal during a detected commercial advertisement segment. Additionally, the system may continue the mute throughout an entire commercial advertisement group containing the detected commercial advertisement segment. To this end, the system determines when the commercial advertisement group began and continues muting for at least some minimum predetermined time period, such as two minutes, following the beginning of the group. In this manner, non-commercial advertisements of the group are muted as well as any commercial advertisements of the group not having a previously stored signature. The manner by which the system determines the beginning of a commercial advertisement group for the purposes of muting the entire group will be described below.

By "muting" it is meant that the amplitudes of the audio and luminance components of the output signal are reduced such that the viewer of monitor 104 sees a dimmed image and hears reduced audio levels. FIG. 3 is a timing diagram illustrating the muting of luminance 132 and audio 134 components of a television signal during two consecutive detected commercial advertisement segments between a pair of un-muted program segments. As can be seen, the levels of both the audio and luminance components are ramped down immediately after the beginning of the first commercial advertisement segment and are ramped up immediately after the end of the second commercial advertisement segment. Mute unit 130 (FIG. 1) also superimposes an audio chime over the audio signal as the audio signal is increased to its pre-mute level. The chime is provided to alert a viewer that the muted commercial advertisement segments have completed. This is particularly desirable if the next program segment begins with a silent audio track, perhaps accompanied by program credits. Without the chime, the viewer, whose attention may be elsewhere, might not realize that a program segment of interest is beginning. In other implementations, the chime may also be produced at the beginning of each muted commercial advertisement segment.

Referring again to FIG. 1, if the signature detected by detector 110 does not match any of the stored signatures of memory unit 128 and the system is not currently muting an entire commercial advertisement group, then no mute control signal is provided to mute unit 130 and the television signal is thereby forwarded without modification by the mute unit to monitor 104. In this manner, program segments remain un-muted. More specifically, when a program segment is received, its signature is detected and compared against those of memory unit 128. Because the memory unit only stores segments corresponding to commercial advertisement segments or other selected broadcast material, the program segment signature will not match any stored segment and no muting will therefore occur during the program segment.

Hence, commercial advertisement detecting and muting system 108 operates to mute in real-time any broadcast segment having a signature matching one of the signatures of memory unit 128. It is, of course, possible that a commercial advertisement will be received for which the signature had not previously been recorded in memory unit 128 and the commercial advertisement will therefore not be muted (unless it is part of an entire group of commercial advertisements being muted). Indeed, in the exemplary implementation, when the system is initially activated, the memory unit has no signatures whatsoever stored therein. Accordingly, detecting and muting system 108 is provided with a non-real-time means for determining whether a broadcast segment, having a signature that does not match any signature recorded in the database, nevertheless is a commercial advertisement segment. Once a determination is made that a broadcast segment is a commercial advertisement segment or other selected segment, the signature of the segment is added to the database such that the segment can be muted in real-time the next time it is encountered.

The determination of whether a broadcast segment is a commercial advertisement segment or other selected segment, assuming that its signature does not match a stored signature, is achieved as follows. As noted above, and as illustrated in FIG. 2, commercial advertisement segments are distinguishable from program segments in that commercial advertisement segments almost always appear in groups of two or more and are each of one or more relatively short pre-determined durations such as 15 seconds, 30 seconds and 60 seconds. Accordingly, if an event marker corresponds to the beginning of a commercial advertisement segment, a second event marker should occur after one of the pre-determined periods of time following the first event marker. If an event marker corresponds to the beginning of a program segment, a second event marker will not likely occur after one of the predetermined periods of time following the first event marker.

Hence, if a second marker occurs, for example, 15, 30 or 60 seconds following a first event marker, a determination is made that the segment between the first and second event markers is therefore probably a commercial advertisement or other selected segment and the signature detected following the first event marker is added to the signatures stored in memory unit 128 (FIG. 1). If a second marker does not occur at 15, 30 or 60 seconds following a first event marker, a determination is made that the segment between the first and second event markers is therefore probably not a commercial advertisement segment and the signature detected following the first event marker is merely discarded.

In practice, for North American broadcast systems, the lengths of commercial advertisements often deviate slightly from the standard 15, 30 and 60 second durations. Accordingly, the system searches for subsequent event markers within a range of time centered around each expected event marker time. For a 30 second commercial advertisement, the system may search, for example, for event markers occurring 28.9 to 32 seconds following a previous event marker.

In any case, if a determination is made that a broadcast segment is a commercial advertisement, then the next time the same commercial advertisement segment is received the following occurs. The signature of the commercial advertisement segment is again detected by detector 110 and compared against those of memory unit 128. The newly detected signature matches the previously stored signature and the commercial advertisement segment is therefore muted in real-time. In this manner, the system builds a database of commercial advertisement segment signatures and each commercial advertisement segment is typically displayed only once in its un-muted form. Thereafter, as long as the signature of the commercial advertisement segment remains in the database, the segment is detected and muted in real-time. Hence, even if the signature database of memory unit 128 is initially empty, a database of commercial advertisement signatures is developed and soon all or most commercial advertisements that are commonly shown are detected and muted. Of course, it is possible for a non-commercial advertisement segment of a desired television program to be 15, 30 or 60 seconds long such that its signature will be added to the database along with those of commercial advertisements. Hence, the next time the same television program is received the segment will be muted under the assumption that it is a commercial advertisement. This is not a particularly significant problem. The viewer merely over-rides the mute and continues to view the program. Moreover, television programs are not typically viewed very frequently and, as will be described below, the system may be configured to erase signatures not frequently encountered such that, the next time the same television program is viewed, the signature of that segment of the program may have already been erased. Also, it is possible for news bulletins, public service announcements etc. to also be 15, 30 or 60 seconds long such that their signatures will be added to the database for future muting. Again, this is not a significant problem. Indeed, if the news bulletins, public service announcements etc. are broadcast frequently, the viewer may consider their subsequent muting to be quite advantageous.

As noted, a determination of whether a segment constitutes a commercial advertisement segment even though its signature does not match a stored signature is performed based upon the time periods separating event markers. FIGS. 4-7 illustrate various examples of event marker groupings within a television signal to help illustrate the manner by which this determination of a commercial advertisement segment is performed.

FIG. 4 illustrates an example wherein a second event marker occurs at one of the pre-determined time periods following a first event marker. More specifically, FIG. 4 shows a pair of event markers 133 and 135 occurring 60 seconds apart. Shortly after first event marker 133 is detected, the signature for the program segment beginning at that event marker is temporarily recorded. Then, a determination is made as to whether a second event marker occurs at 15, 30 or 60 seconds following the first event marker. In the example of FIG. 4, second marker 135 is detected 60 seconds from the first marker. Accordingly, shortly thereafter, the temporarily recorded signature is saved i.e. it is stored in memory unit 128 (FIG. 1) along with any other pertinent information such as the duration of the corresponding commercial advertisement as determined by the time period between the first and second event markers. If the second marker had alternatively occurred at either 15 or 30 seconds following the first event marker, the signature would have been saved then instead.

With further reference to FIG. 4, it should be noted that the detection of second event marker 135 also triggers a temporary recording of the signature of the segment beginning at the second event marker. This signature is compared with those stored in the memory unit (FIG. 1). As with the first signature, if a match is detected then the broadcast segment beginning at the second event marker is immediately muted and if no match is detected, then the signature is temporarily recorded until a determination can be made whether the broadcast segment beginning at the second event marker is nevertheless a commercial advertisement segment.

FIG. 5 illustrates an example wherein a second event marker does not occur at one of the pre-determined time periods following a first event marker. More specifically, FIG. 5 shows a pair of event markers 136 and 138 occurring more than 60 seconds apart. Shortly after first event marker 136 is detected, the signature for the broadcast segment beginning at that event marker is temporarily recorded. Then, a determination is made as to whether a second event marker occurs at 15, 30 or 60 seconds following the first event marker. In the example of FIG. 5, no event marker is detected at either 15, 30 or 60 seconds following the first event marker. Accordingly, shortly after the time of event marker 136 plus 60 seconds, the temporarily recorded signature is discarded. (Again it should be noted that the detection of the second event marker also triggers a temporary recording of the signature of the segment beginning at the second event marker and processing of that signature proceeds.)

FIG. 6 illustrates an example where a pair of first and second event markers 140 and 142 are separated by less than 60 seconds but where second event marker 142 does not occur at one of the predetermined intervals. Hence, the first marker does not correspond to the beginning of a commercial advertisement or other selected segment. It should be noted that the determination that the first marker does not correspond to a commercial advertisement is not made until 60 seconds following the detection of the first event marker. Accordingly, the temporarily recorded signature recorded following the first event marker is not discarded until after the 60 second mark. The detection of the second event marker in the interim does not terminate the search for additional event markers based upon the first event marker. That search continues until at least 60 seconds from the first event marker. Also, the detection of the second event marker triggers a temporary recording of a second signature and a contemporaneous search for additional event markers following the second event marker at 15, 30 or 60 second intervals.

FIG. 7 illustrates an example where a pair of event markers 144 and 146 are separated by 30 seconds. Hence, the first marker corresponds to a commercial advertisement, and a signature temporarily recorded shortly after the first event marker is stored along with other pertinent information in the database of memory unit 128 (FIG. 1). It should be noted that the determination that the first and second event markers 144 and 146 correspond to the beginning and end of a commercial advertisement is made shortly after second event marker 146 is detected and is not deferred until 60 seconds from the first event marker. Also, FIG. 7 shows another event marker 148 occurring between event markers 144 and 146 but not at one of the predetermined time periods. The detection of the additional event marker in the interim between event markers 144 and 146 does not terminate the search for an additional event marker at any of the predetermined time periods from the first event marker. The detection of event marker 148 however triggers a temporary recording of another signature immediately thereafter and a contemporaneous search for event markers following at 15, 30 or 60 second intervals.

Hence, FIGS. 6-7 illustrate that more than one signature may be temporarily recorded at any given time and that multiple contemporaneous searches may be performed for subsequent event markers with one search for each event marker. Indeed, in the exemplary implementation, each event marker, for which the corresponding signature is not found in the signature database, triggers a search for an additional event marker at one of the pre-determined intervals thereafter. As each search may last as long as 60 seconds during which time a number of additional event markers may be encountered, a number of signatures may need to be simultaneously temporarily stored and coordinated. Additionally, the time that each event marker is detected is recorded to allow calculation of the various time intervals. Accordingly, a temporary signature and event marker time memory unit 150 is provided in connection with control unit 126 of FIG. 1 for storing the temporary signatures and the time of the event marker corresponding to each temporarily stored signature. Signatures stored therein are either erased after 60 seconds, if no event marker is detected at any of the pre-determined intervals, or transferred to the signature memory unit 128, if an event marker is detected at any of the pre-determined intervals. In the exemplary embodiment, control unit 126 of FIG. 1 coordinates storage of information within temporary memory unit 150 and performs the various searches for additional event markers. In this regard, the control unit may be a microprocessor pre-programmed with appropriate software to coordinate the aforementioned operations. Alternatively, the control unit may be a dedicated integrated circuit pre-configured with appropriate hard-wired logic to coordinate the operations. Other appropriate devices may be employed as well.

Thus, by detecting various event marker times, the system can determine whether a segment, which does not have a matching stored signature, is nevertheless a commercial advertisement segment. Hence, each received commercial advertisement is detected either immediately via a signature match or at the end of the segment via an event marker analysis. Once a commercial advertisement is detected, the system determines whether it represents the beginning of a new commercial advertisement group and, if so, muting is maintained for at least two minutes thereafter, regardless of whether the subsequent segments have matching signatures in the database. In one implementation, this is achieved as follows. For any detected commercial advertisement segment, if no other commercial advertisement segment preceded it during the previous 60 seconds, or some other predetermined minimum period of time, then the system identifies the detected commercial advertisement segment as being the first of a new group. If it were not the first of a new group, then a previous commercial advertisement segment of the same group would at least have terminated at some point during the preceding 60 seconds. (This assumes that no two consecutive commercial advertisement segments of a common group are ever separated by more than 60 seconds of noncommercial advertisement segments such as station breaks etc. In most cases, this assumption is correct.) Once the system identifies the first commercial advertisement of the group, muting is initiated and maintained for at least two minutes, or some other predetermined minimum group time period, from the beginning of that commercial advertisement Notably, this determination is made regardless of whether the commercial advertisement was detected via signature match or event marker analysis. If detected by signature match, the determination is made during the reception of the commercial advertisement. If detected via event marker analysis, the determination is not made until the end of the segment. In either case, the duration of continuous muting is calculated based on the beginning of the first commercial advertisement segment of the group. As noted, muting is maintained for some predetermined period of time representative of the minimum duration of a group of commercial advertisements such as two-minutes. Any particular group may be longer in duration and continuous group-based muting may therefore end during the group. Signature-based muting, however, will continue for any remaining commercial advertisements of the group that have matching signatures in the database. Hence, the remaining advertisements of the group will not necessarily be unmuted. Other techniques for detecting the beginnings and predicting the endings of commercial advertisement groups may alternatively be employed.

With reference to the remaining Figures, further details of exemplary and alternative method and apparatus embodiments of the invention will be described.

FIG. 8 is a flowchart summarizing steps performed by the commercial detection and muting system of FIG. 1 or by any other appropriate system for performing commercial advertisement detection and muting. As many of these steps have already been describe above, only additional aspects of the detection and muting technique will be described in detail. Initially at step 200, a television broadcast signal is received. The television signal is searched to detect event markers, step 202. If an event marker is not detected, step 204, the television signal is merely output at step 206 perhaps to a display device such as monitor 104 (FIG. 1). Steps 200, 202, 204 and 206 are repeated until an event marker is detected at step 204 triggering the beginning of a broadcast segment evaluation process for the broadcast segment beginning at the detected event marker, step 208. As will be described below, the broadcast segment evaluation process operates to determine whether the broadcast segment beginning at the corresponding event marker is a commercial advertisement and if so mutes the output of the television signal during the commercial advertisement. After the broadcast segment evaluation process has been triggered for that event marker, step 206 is again executed to continue outputting the television signal. Hence, the television signal is not stored or otherwise delayed but is immediately output in real-time. The broadcast segment signature evaluation process is performed shortly after detection of the event marker such that very little of the television signal is output before a determination of whether the signal needs to be muted is achieved.

Steps 200-206 are repeatedly performed to search the television signal for additional event markers. Each newly detected event marker triggers a separate broadcast segment signature evaluation process for the broadcast segment beginning at that event marker. If part of the system performing the method of FIG. 8 is implemented as a software program, the individual broadcast segment evaluation processes may be, for example, separate and distinct software processes running on a microprocessor or may merely be separate logic threads within a single software process. They are described herein as separate processes primarily to help clarify that contemporaneous searches for subsequent event markers may need to be performed for each of a set of consecutive event markers and not to imply any particular architecture by which the method of FIG. 8 must be implemented.

Beginning at step 210, one broadcast segment evaluation process triggered at step 208 is set forth. Step 210 is shown connected to step 208 by a dashed line to indicate that the broadcast segment evaluation process proceeds independently of, and contemporaneously with, steps 200-206. At step 212, the signature corresponding to the current broadcast segment is read from the television signal. Additional details of step 212 are step forth in FIGS. 9-11 and will be described below. At step 214, the signature is compared against stored signatures in a database of signatures known to correspond to commercial advertisements or other unwanted broadcast material. If there is no signature match, step 216, then step 218 is performed to determine whether the current broadcast segment nevertheless comprises a commercial advertisement segment and, if so, what the length of the commercial advertisement segment is in seconds. Additional details of step 218 are set forth in FIG. 12 and will be described below. If the current segment is determined to not be a commercial advertisement at step 220, then step 222 is executed to end the broadcast segment evaluation process. If, however, the current segment is determined to be a commercial advertisement, step 224 is executed to store the signature of the current segment read at step 212 in the database such that, when the same commercial advertisement segment is encountered again, the commercial advertisement can be detected in real-time. As noted above, the determination of whether a broadcast segment comprises a commercial advertisement segment may take up to 60 seconds from the event marker defining the beginning of the segment. Additional details of step 224 are set forth in FIGS. 13-16 and will be described below.

At step 224, in addition to storing the signature for the current broadcast segment, the system also stores the length of the commercial advertisement and other pertinent information such as the latest date and time of day (date/time) of the broadcast of the segment. The length is used to determine how long to mute the output signal if the same commercial advertisement is again encountered. The latest date/time of broadcast can be employed to selectively erase signatures from the database based upon how recently the signatures had been encountered. More information regarding the content and purpose of information stored in the database along with the signature will be described in greater detail below. After the database is updated at step 224, step 225 is performed to determine whether the detected commercial advertisement is the first of a group and if so continuous group-based muting, described above, is performed until at least the end of the minimum pre-determined group time period, assuming no user over-ride. Then, step 222 is executed to end the current broadcast segment evaluation process.

Hence steps beginning at step 218 are performed only if the signature of the current broadcast segment does not match any in the database as determined at step 216. If there is a signature match at step 216 indicating that the current broadcast segment is a commercial advertisement, the system first determines whether the commercial advertisement is the first of a group, step 226, and then begins muting the output television signal, step 227, by using a mute control signal in the manner described above with reference to FIGS. 1 and 3. Accordingly, television signals output at step 208 are muted and will remain muted until the mute control signal is deactivated. If the advertisement is the first of a group, the muting will ultimately continue until the end of the predetermined minimum group time period as described above, assuming no user over-ride. If not, muting will continue until the end of the advertisement, again assuming no over-ride. The length of the commercial advertisement is then read from the database, step 228, such that system can determine the end of the current commercial advertisement segment. Hence, if a previously triggered group-based muting terminates during the current segment, the system will continue to mute until the end of the segment. At step 230, the system updates the parameters stored with the signature, such as the latest date and time of broadcast. Then, at step 232, the muting is terminated either upon user override or upon the latter of the completion of the current commercial advertisement segment and the completion of the current group (as determined by the minimum predetermined group time period) if group-based muting had been triggered. The current commercial advertisement segment, as with all segments, ends with an event marker which is detected at step 204 triggering another broadcast segment evaluation process.

Hence, FIG. 8 illustrates that broadcast segment evaluation processes are triggered upon the detection of each event marker. As discussed above, each broadcast segment evaluation process may take up to 60 seconds to determine whether the broadcast segment beginning at the event marker triggering the process is a commercial advertisement or not. During that period of time a number of other event markers may be detected and a broadcast segment evaluation process triggered for each one. Hence, a number of evaluation processes may proceed simultaneously. The resolution of one process may moot or obviate some of the latter-triggered processes. For example, if a first triggered process determines that the corresponding broadcast segment is a commercial advertisement, then any event markers detected before the end of the broadcast segment thereby occurred during the commercial advertisement and therefore do not represent the beginnings of unique commercial advertisements, and the broadcast segment evaluation processes triggered thereby may be terminated. In other words, if a commercial advertisement is detected because a pair of event markers are found to be exactly 60 seconds apart, any other event markers detected during that 60 second period thereby occurred during the 60 second commercial and therefore can be ignored. On the other hand, if after 60 seconds a first triggered process determines that a first event marker at the beginning of that 60 second period is not the beginning of a commercial advertisement, then any event marker detected after that first event marker may itself represent the beginning of a commercial advertisement segment, and the broadcast segment evaluation processes triggered thereby may not be terminated until the situation is further resolved. Conversely, if a second triggered process determines that the corresponding broadcast segment is a commercial advertisement, then any event markers detected before the beginning of that broadcast segment occurred before the commercial advertisement began and therefore do not themselves represent the beginnings of unique commercial advertisements, and any broadcast segment evaluation processes triggered thereby may be terminated. For clarity in illustrating the basic process of the invention, the details of the logic by which the resolution of one broadcast segment evaluation process affects others are not set forth in FIG. 8 but may be readily incorporated within a computer program or other logic device configured for implementing the method of the invention in accordance with the teachings provided herein.

With reference to FIGS. 9-16 details of exemplary implementations of some of the steps of FIG. 8 will now be described.

As noted, step 212 of FIG. 8 operates to read a signature corresponding to an event marker from the broadcast signal. Any of a wide variety of techniques for defining the signature of a broadcast segment and for reading and recording the signature may be employed. FIG. 9 sets forth one exemplary technique for reading a signature corresponding to an event marker from a broadcast signal. At step 236, the system waits a pre-determined number of frames from the completion of the event marker. The system may wait, for example, 10 frames or the equivalent period of time. At step 238, the system then records the current frame, e.g., the system takes a snap shot of the tenth frame, hereinafter referred to as the test frame. An exemplary test frame 240 is illustrated in FIG. 10. At step 242, the system determines, for each of a preselected set of signature test lines of the test frame, the average luminance level of the lines. The signature test lines may be, for example, the first 64 odd lines of the frame beginning at line 23 or perhaps the 5th, 10th, 15th etc. lines. A few exemplary signature test lines are identified in FIG. 10. The average luminance level of each line may be determined, for example, by routing the luminance signals for those lines through circuitry (not shown) configured to generate a test voltage level representative of the average voltage level of the luminance signal. At step 244 a determination is made, for each signature test line, as to whether the average luminance level for that line exceeds a pre-determined threshold. This determination may be made, for example, by defining the threshold level in terms of a threshold voltage level and comparing the test voltage level with the threshold voltage level. At step 246, the system assigns a bit value of 1 to each line exceeding the threshold luminance level and a bit value of 0 otherwise. Of course, opposite bit values could alternatively be employed. At step 248, the bit values are concatenated together yielding a single binary bit string hash code signature with one bit per test line of the test frame. An exemplary bit string 250 is set forth in FIG. 11. Execution then returns to step 212 of FIG. 8 via step 249 where the signature is compared against those in the database.

Hence, a bit string signature is evaluated based upon the average luminance level of selected lines of one selected frame following the event marker. Any bit string of suitable length may be employed. If too few test signature lines are selected yielding a short bit string, the bit string may not have sufficient bits to uniquely identify the broadcast segment. Accordingly, it is preferred that at least 32 signature test lines are selected yielding a bit string of at least 32 bits. With 32 bits, it is reasonably unlikely that test frames from different program segments will have the same bit string and hence, the bit string can be used to uniquely identify each unique broadcast segment. Also, a test frame occurring fairly promptly after the event marker should be selected. If a test frame is selected that occurs too late after the event marker, the system will not be able to promptly determine whether the signature of the test frame matches any of the recorded signatures and will therefore not be able to promptly begin real-time muting. Hence, it is preferred that the test frame occur no longer than about one second after the event marker. In alternative implementations, signature test lines may be drawn from two or more separate test frames.

Thus FIGS. 9-11 set forth exemplary techniques for reading a signature wherein a bit string is generated based upon whether average luminance levels of selected signature test lines of a selected test frame exceed a threshold level. Of course, the same test signature lines, the same test frame and same threshold level must be employed each time to ensure that the same bit string is generated every time the same broadcast segment is encountered. Hence, care should be taken in detecting and comparing signatures to compensate for skew or other factors in the television signal. Details of one possible system for detecting and comparing signatures which addresses skew and other factors is set forth in U.S. Pat. No. 5,504,518 summarized above.

If the television signal is digital, average luminance levels need not be employed. Rather, the signature is simply based on selected bits within one or more selected digital frames. Skew and other related problems are also avoided or more easily overcome.

As noted, step 218 of FIG. 8 operates to determine if the current broadcast segment nevertheless corresponds to the beginning of a commercial advertisement even though the signature of the segment beginning with the event marker triggering the evaluation process does not match any in the database. FIG. 12 sets forth one exemplary technique for making the determination. An overview of the technique was provided above in connection with FIGS. 4-7. At step 252 a set of predetermined time periods based on standard commercial advertisement lengths are read from a memory. Examples include 15 seconds, 30 seconds, 60 seconds. At step 254, the system sets appropriate timers to begin waiting each of the pre-determined time periods. At step 256, the system determines whether a second event marker occurs at any of the predetermined time periods following the first event marker. For example, the system first waits 15 seconds from the time of the first event marker and detects whether the television signal includes an event marker at that point in time. If not, the system waits until 30 seconds from the first event marker and so on. If a second event marker is detected at any of the predetermined times, step 258, then step 260 is performed wherein the broadcast segment beginning at the first event marker is identified as a commercial advertisement. At step 262, the system determines the length of the commercial advertisement by comparing the time the second event marker was detected with that of the first marker. Execution thereafter returns through step 264 to step 218 of FIG. 8. If at step 258, no event marker is found at any of the pre-determined time periods following the first event marker, then step 266 is performed wherein the broadcast segment starting at the first event marker is identified as not being a commercial advertisement segment and again execution returns to step 218 of FIG. 8.

Step 224 of FIG. 8 operates to store the signature and other pertinent information in the database if the broadcast segment is found to be a commercial advertisement. The database, however, may be full. A determination then needs to be made as to which previously stored signatures, if any, should be overwritten to make room for the newly recorded signature. FIGS. 13-16 set forth details of one exemplary technique. At step 268, the system determines whether the database is full and if not, step 270 is performed to store the signature, the length of the commercial advertisement for the segment corresponding to the signature and the date/time the event marker marking the beginning of the commercial advertisement was detected. Other information may also be recorded for the purpose of allowing the system to intelligently decide which records to erase as the database becomes full. For example, if the system is configured to detect commercial advertisements even while the television set is otherwise not in use (as will be described in greater detail below with reference to FIG. 17), the system also stores an in-use weighting factor, which may be a single bit value, indicating whether the commercial advertisement was detected during a period of time when the television was not otherwise in use or whether it was detected while the television was in use. The system may also increment a counter identifying the number of times the commercial advertisement is encountered.

FIG. 14 sets forth an exemplary database (for use as signature memory unit 128 of FIG. 1) having, for each unique signature stored therein, entries for the first broadcast date/time, the latest broadcast date/time, the number of times the commercial advertisement is encountered and the in-use weighting factor. When the signature entry is initially recorded (step 224 of FIG. 8), the first date/time and the latest date/time are the same, the number of broadcasts is one and the in-use weighting factor is either set to 1 or 0 depending upon whether the television set is currently in use.

Within FIG. 13, if the database is found to be full at step 268 then step 272 is performed wherein the system selects one of the previously stored records in the database to be overwritten based upon, for example, the latest date/time that the previous signatures were encountered, the first date/time the signatures were encountered, etc. For example, depending upon system pre-programming, the system may overwrite whichever entry was least recently written to the database as indicated by the latest date/time of broadcast information (i.e., first in first out). This technique has the advantage of overwriting signatures corresponding to commercial advertisements not encountered recently and thereby keeping the database up to date. If such is the case, no further information need be recorded along with the signature other than the latest date/time which must be updated upon each detection of the signature (see step 230 of FIG. 8). As another example, the system may overwrite whichever signature record has the lowest frequency of occurrence. In such a case, the system calculates a frequency of occurrence based upon the first date/time of entry, the latest date/time of entry and the number of broadcasts encountered. This technique has the advantage of overwriting signatures not frequently encountered even though the signature may have been recently encountered.

As another example, the system may overwrite whichever signature simply has the least number of encountered broadcasts as represented by the number of broadcasts counter. To ensure that newly stored signatures are not immediately overwritten the system may also look at the earliest time of broadcast and overwrite a signature only if it was recorded some minimum amount of time earlier.

If the system is configured to search for commercial advertisements even while the television set is not otherwise in use, the system may be configured to overwrite signatures encountered during times that the television was not otherwise in use first before overwriting signatures encountered while the television set was in use. This has the advantage of eliminating signatures that correspond to commercial advertisements not ordinarily encountered while the viewer of the television set typically watches television. In this regard, whenever a signature is first encountered, the system simply determines whether the television set is on and if so, sets the in-use bit to 1 if the set is on and 0 otherwise. Thereafter, if the signature is encountered while the in-use bit is set to 0 and while the television set is on, the in-use bit is re-set to 1. Once the in-use bit is set to 1 it is not reset to 0 until the record is erased. Hence, the in-use bit indicates whether the signature has been encountered at least once while the television set was on. In any case, if a signature record needs to be overwritten, records having in-use bits set to 0 are overwritten first before any records having in-use bits is set to 1.

In yet another example, the system may overwrite whichever entry was first written to the database as indicated by the first date/time of broadcast information. If such is the case, the memory is preferably implemented, as illustrated in FIG. 15, as a FIFO 274 such that no explicit date/time value need be stored. If implemented as set forth in FIG. 14, the memory is preferably a RAM.

In yet another implementation, illustrated in FIG. 16, the database is capable of storing multiple records per signature. More specifically, database 276 of FIG. 16 stores a record identifying the broadcast date/time of each detection of each commercial advertisement signature. Accordingly the database maintains a record of when each commercial advertisement is broadcast. Exemplary signatures #1.sub.1 -#1.sub.N and #2.sub.1 -#2.sub.M are shown. Periodically, the contents of the database are output to provide a list, perhaps for broadcast verification purposes, of the date and time of each broadcast of each commercial advertisement. Preferably, the database is of sufficient size such that records do not need to be overwritten prior to outputting the date/time broadcast information.

As noted above, in one implementation the system is configured to operate in a background mode to detect commercial advertisements and record the corresponding signatures even while the television set is not otherwise in use. This is provided in part to allow the system to quickly build a database of commercial advertisement signatures such that, when the television set is in use, there is a higher probability that commercial advertisements encountered will be muted. The background mode is particularly advantageous for use when the system is initially activated with few, if any, pre-stored signatures.

FIG. 17 sets forth a method for implementing the background mode. Initially, at step 300, commercial detecting and muting system 108 (FIG. 1) is powered-up with audio and video outputs to television monitor 104 (also FIG. 1) disabled. At step 302, the database is examined to determine whether it is complete. Depending upon preprogramming, the database may be regarded as being complete if it is full or perhaps if it holds some threshold number of signature records. Assuming that the database is complete, then step 304 is performed where the system waits for user input to turn the television monitor on. Hence, no background processing is initiated. Once the monitor is activated, the audio and video are enabled, step 306, and the system begins receiving a broadcast signal, step 308. As with the basic method of FIG. 8, the system searches for event markers, step 310, and if event markers are detected, step 312, the system begins a broadcast segment evaluation process to determine whether the broadcast segment is a commercial advertisement, step 314. Details of the broadcast segment evaluation process are set forth in FIGS. 8-16 and have been described above. Then, regardless of whether an event marker has been detected or not, step 316 is performed to output the received broadcast signals to the monitor. Since audio and video had previously been enabled, the broadcast signal is thereby displayed by the television monitor, although perhaps muted if currently displaying a detected commercial advertisement. At step 318, the system determines whether the user has de-activated the television monitor and, if not, steps 308-318 are again performed. In this manner, steps 308-318 are repeated in a loop until the television monitor is turned off--at which time step 320 is performed to again disable audio and video. At step 302, the database is again examined to determine if it is still complete and if so the system waits at step 304 until the user again activates the television monitor. Step 302 is performed following step 320 because, in some implementations, the user is provided with the capability of clearing all or a part of the commercial advertisement signature database while operating the television set. Hence, the commercial advertisement database may no longer be complete. If at step 302, the database is not complete, either because the system is newly installed and has not yet built up a complete database or because the user has cleared the database, execution proceeds to step 322 where the system determines whether the user has activated the television monitor and if so execution proceeds to step 306 and following for non-background commercial advertisement detection processing as already described. If the user has not activated the television monitor, execution instead proceeds immediately to step 324 where background processing begins. (It should be noted that step 322 differs from step 304 where the system waits for the user to activate the television monitor. With step 322, the system does not wait for the user to activate the television monitor but immediately triggers either background or non-background processing.)

Background processing begins at step 324 wherein broadcast signals are received. As with the basic method of FIG. 8, the system searches for event markers, step 326, and if event markers are detected, step 328, the system begins a broadcast segment evaluation process to determine whether the broadcast segment is a commercial advertisement, step 330. Details of the broadcast segment evaluation process are set forth in FIGS. 8-16 and have been described above. Then, regardless of whether an event marker has been detected or not, execution returns to step 302 to again evaluate whether the database is complete. It should be noted that in the background processing loop of steps 302-330, there is no step of outputting the broadcast signal. Since audio and video had previously been disabled, no output signal is needed.

If the database remains incomplete and the television monitor remains deactivated, the background processing mode of steps 302-330 is performed in a loop to build a signature database. In this regard, every time a broadcast segment evaluation process is triggered at step 328 which detects a commercial advertisement, the signature for the advertisement is added to the database in the manner described above with reference to FIGS. 8-16. The background processing loop ends either when the database becomes complete (at which time the system waits at step 304 for user input to activate the television monitor) or when the user activates the television monitor at step 322 even though the database is not yet complete. In either case, non-background processing is then performed.

Thus FIG. 17 sets forth a method for operating the system in a background mode to build a signature database even while the television set is not otherwise in use and further sets forth a technique for determining whether background or non-background processing should be performed based upon the completion state of the database and upon whether the user activates the television monitor. In other implementations, rather than detect whether the database is full and activate the background processing mode if it is not full and the user has not activated the monitor, the system merely operates in background mode for some predetermined period of time after the initial system power up (unless the user has activated the monitor). The predetermined period of time may be, for example, two weeks.

As noted, a system employing the above-described techniques may be implemented either as a plug-in module or as an external device. If a plug-in module, the module is interconnected between an output of a tuner portion of the television set and an input to an audio/video display portion of the television set to output selectively muted or otherwise altered audio and video signals. If external, the system may be interconnected to a television set as shown in FIG. 18.

A selective muting unit 400 is connected to audio 402 and video 404 output jacks of a television set 406. The audio and video output jacks are connected internally to a tuner portion (not separately shown) of the television set. Selective muting unit 400 receives the audio and video signals from the tuner via the audio and video output jacks and determines, using the techniques described above, when to initiate and terminate muting. An infrared muting transmitter 408 receives an electrical muting signal from muting unit 400 and outputs an infrared signal to an infrared input 410 of the television set. The infrared muting signal is set to emulate the infrared signal normally received by the television set from an infrared remote control unit (not separately shown) when a user presses a mute button on the remote control. In this manner, the audio is muted during detected commercial advertisements or other unwanted broadcast material as described above. The mute signal is transmitted at the beginning of a detected commercial advertisement and a second un-mute signal is transmitted either at the end of the commercial advertisement or at the end of a group of commercial advertisements.

If the television set is provided with a video mute capability triggered by a video mute infrared signal from a remote control unit, then infrared muting transmitter 408 additionally transmits the video mute signal to thereby provide both video and audio muting. Other television functions responsive to infrared input control may be additionally or alternatively exploited. Also, if the television set is provided with audio and video input jacks along with the audio and video output jacks, muting unit 400 may be configured to output muted audio and video signals into the audio and video input jacks of the television set. In such an embodiment, the television set is set to a mode for outputting signals received via the tuner of the television set through the audio and video output jacks and for displaying signals received through the audio and video input jacks. Hence, selectively muted television signals are display in the same manner as if the muting unit were mounted internally within the television set.

Thus far, the invention has been described with respect to exemplary embodiments employing television sets. FIG. 19 illustrates an embodiment employing a PCTV 500 configured to operate either in a television mode or a personal computer (PC) mode. Depending upon the mode, PCTV 500 receives television signals via a cable 502 and an internal tuner (not separately shown) or receives and transmits modulated data signals over a telephone line 504, or other data line such as an ISDN line. Alternately, both television and modulated data input and output may be via a single line such as a cable line. In any case, while in television mode, PCTV 500 displays television signals via a monitor 506 and plays audio signals via speakers 508. While in PC mode, the PCTV displays graphic images generated by computer programs such as web browsers, word processors, spread sheets and the like via monitor 506, plays any corresponding audio via speakers 508 and receives user input via a keyboard 510.

PCTV 500 is provided with hardware or software or both (not separately shown) for selectively muting the audio and video of television signals while operating in television mode in the same manner as described above. Hence, commercial advertisements or other unwanted broadcast material are selectively muted. When implemented via hardware, a plug-in module as described above is provided. The plug-in module either has its own memory for storing signatures and other information or is configured to employ a disk drive or other storage unit 512 of the PCTV for such purposes. Signatures may be stored temporarily during processing with a RAM (not shown) of the PCTV to expedite processing. If implemented in software, the software may, for example, run on a microprocessor of the PCTV to process digitized portions of the television signal (provided by an analog to digital converter not separately shown) for muting purposes, again using a disk drive or other storage unit 512 for storing signatures and the like. If the PCTV is configured to receive digitized television signals, then no tuner or analog to digital converter may be needed.

When in PC mode, the PCTV does not operate to perform any muting, unless the PC simultaneously displays the television signals within a window of monitor 506. Principles of the invention may be additionally applied to detect and selectively process other types of incoming data signals besides television signals to mute or otherwise alter unwanted portions of the signals, including signals otherwise processed only in PC mode including, for example, advertisement portions of input web pages.

What has been described is a method and apparatus for detecting commercial advertisements by comparing signatures of received broadcast segments with a database of stored signatures corresponding to known commercial advertisements and for muting any segments having a signature matching any of the stored signatures in real-time. For segments having signatures not found in the database, the system nevertheless determines whether the segment is a commercial advertisement and, if so, adds the signature to the database. This latter determination is based upon known characteristics of commercial advertisements. In the exemplary embodiment, the determination is based upon the assumption that commercial advertisements are always preceded and followed by black frame/low audio event markers and are always 15, 30 or 60 seconds in length. This appears to be a fairly universal rule at least with television signals broadcast in the United States. However, if employed in connection with a television broadcast system wherein commercial advertisements are typically of different lengths, the system of the invention may be pre-programmed with those different lengths. In other implementations, other appropriate rules or characteristics of commercial advertisements or other selected segments may employed either additionally or alternatively for distinguishing the selected segments from other broadcast segments.

In general, almost any set of rules or characteristics of broadcast signals that serve to distinguish commercial advertisements or other selected segments from program segments may be employed to determine whether or not a broadcast segment, for which there is no signature match, is nevertheless a commercial advertisement. The rules are merely programmed or otherwise incorporated within system 108 of FIG. 1 to detect commercial advertisements and an appropriate signature or other distinguishing feature is stored for application against future received broadcast signals. Indeed, suitable rules may be applied to detect almost any desired selected segment within a television broadcast signal and not necessarily just commercial advertisements. Moreover, principles of the invention may be applied to signals other than television signals such as to radio broadcast signals.

Moreover, although described with respect to television sets or PCTV's, any device or computer configured to receive television or other broadcast signals can be provided with appropriate hardware or software or both for detecting and selectively processing portions of the broadcast signals in the manner described above.

Although described with reference to an exemplary system which operates to mute a television signal during commercial advertisements, almost any other desired action may alternatively be triggered. For example, the system may completely cut off the broadcast video and audio feed and replace the video feed with, for example, a screen presenting an appropriate logo representative of the commercial determination system such as the tradename or trademark of the system, perhaps Sound!Off.TM.. As another example, the system may change the reception channel upon detection of a commercial advertisement then return to the same channel upon completion of the advertisement or group of advertisements. Such requires a two-tuner system. Also, as described above, the system may merely record the date and time of each commercial advertisement for broadcast verification purposes. As can be appreciated, a wide range of alternative implementations are available consistent with the general principles of the invention.

The preceding description of exemplary embodiments is provided to enable any person skilled in the art to make or use the invention. It should be appreciated that not all components necessary for a complete implementation of a practical system are illustrated or described in detail. Rather, only those components necessary for a thorough understanding of the invention have been illustrated and described. Furthermore, it should be understood that the exemplary embodiments of the invention described herein are merely illustrative of general principles of the invention which can be applied to other applications, to achieve other ends, and to remedy other problems not specified herein. Accordingly, the scope of the invention should not be limited to the exemplary embodiments described herein.