Television – Format – Including additional information
Reexamination Certificate
1997-03-28
2001-04-03
Peng, John (Department: 2614)
Television
Format
Including additional information
C348S461000, C348S475000
Reexamination Certificate
active
06211919
ABSTRACT:
CROSS-REFERENCE TO RELATED APPLICATIONS
Not applicable
STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT
Not applicable
BACKGROUND OF THE INVENTION
The present invention relates to data transmission, and more particularly to a method and apparatus for transparently embedding data within a video signal in order to indicate authentication, program ownership and/or reception-verification of the video signal.
Television signals are usually copyrighted or otherwise proprietary to the originator and, in the case of television network distribution, are distributed to affiliate stations for re-broadcast. Unauthorized re-broadcast is difficult to detect since it is often difficult to determine the creator or originator of the material from the material itself. This is particularly true for short sequences or when the video image has been cropped or modified to mask any identification logos that may have been explicitly burned into the video image.
Another common problem with television and video distribution, which is becoming more severe, is the synchronization of an aural or audio component that is to be distributed with the video signal. Modern digital signal processing techniques requiring large buffering, such as MPEG compression, add latency to the video distribution and, since the audio may be distributed or processed separately, an error in audio to video “lip-sync” or in sound to action often occurs. Sometimes this latency is variable, requiring continual re-synchronization. In addition to audio to video synchronization problems, the audio signal may be distorted or missing entirely. By embedding, for example, the audio envelope as data within the video signal, it is also possible to detect and compare the received audio with the original which was coded and embedded as data within the video signal as a quality measure as well as video to audio delay.
Another problem is with authentication of video signals representing visual images, possibly created by a computer, where it is desirable to detect a forgery or imposter signal. In this case the video signal may be replicated to such a degree that it is difficult to detect it from the authentic video sequence. Also it is sometimes desirable to detect the beginning and/or ending of a particular sub-sequence of a video signal, such as a motion test sequence for in-service video quality assessment of video processed by MPEG compression. It is possible to devise a remote receiver to capture a special test sequence segment of a distributed video signal and compare that segment to an undistorted, stored version to assess the quality of the video. Although it is possible to uniquely and reliably detect when the sequence occurs so that it, and only it, is captured, this generally requires the advance processing of a signature vector, transmission of that vector through a separate channel and a correlation means at the receiver. By transparently embedding data into the video a predetermined start and end code may be detected by the receiver without any preloading of signature vectors and signature vector preprocessing.
Published Canadian Patent Application No. 2,174,413 (A1) by Geoffrey B. Rhoads entitled “Identification/Authentication Coding Method and Apparatus” discusses techniques for providing authentication of image signals. In Rhoads an imperceptible N-bit identification code is embedded throughout an image with a small noise pattern in a coded fashion. In particular bits of a binary identification code are referenced sequentially to add up to N independent, noise-like patterns to the original image signal. The detection of these patterns is done by N sequential correlations with stored replicas of each pattern. This may also be done simultaneously by N correlators, as is well described in the public domain as a “correlation receiver.” Rhoads further discloses adding or subtracting exactly N independent, noise-like images to improve detection and/or encoded image quality. This later modification, referred to by Rhoads as “true polarity”, is well described in the public domain as “bi-orthogonal signaling” for a correlation receiver allowing 2{circumflex over ( )}N composite symbols (patterns) to be created by adding or subtracting N bi-orthogonal symbols (patterns). A disadvantage of both of Roads' methods is that each of the N noise-like patterns, which are added or subtracted to form one of the 2{circumflex over ( )}N possible composite patterns, needs to be properly scaled and designed to minimize image degradation by the addition of the composite pattern to the image. A further disadvantage is that the original unencoded image, as well as the N patterns, need to be stored in the receiver so that the unencoded image may be subtracted from the encoded image for detection.
U.S. Pat. No. 4,969,041 issued Nov. 6, 1990 to William J. O'Grady and Robert J. Dubner entitled “Embedment of Data in a Video Signal” discloses adding one of a plurality of low-level waveforms to the video signal, with the level of the low level waveform being below the noise level of the video signal, for authentication purposes or for transmitting information. Each low-level waveform corresponds to a particular data word being embedded. At the receiving end the video signal is correlated with an identical set of low-level waveforms to produce a set of correlation coefficients—the highest correlation coefficient indicating the presence of a particular one of the low-level waveforms which is then converted into the corresponding data word. The Rhoads composite pattern consists of the summation of up to N independent patterns, which would be the same as O'Grady/Dubner when N=1 since only one pattern is sent at a time. For an N-bit data word O'Grady/Dubner implies the need for storage of 2{circumflex over ( )}N patterns rather than the N patterns of Rhoads. But since 2{circumflex over ( )}N patterns may be generated by summing N bi-orthogonal sub-patterns, the effect is the same and only N patterns need to be stored in a manner identical to Rhoads. In both of these patents the symbol rate is one pattern per picture or field.
Some other drawbacks of O'Grady/Dubner are that it limits the degree to which a pattern or sequence of patterns can be hidden by not fully exploiting the limitations of the human psychovisual process, such as spatial masking; it does not fully exploit the available signal to noise ratio since the correlation output is unipolar; and it does not include coding of the data represented by the embedded symbols so as to provide temporal redundancy for error correction over missing video frames.
What is desired is a method of embedding an unobtrusive data pattern into the video signal that is useful in addressing the above-noted problems.
BRIEF SUMMARY OF THE INVENTION
Accordingly the present invention provides a method and apparatus for transparently embedding data into a video signal by modulating a particular carrier frequency with a randomized pattern representing the data. The randomized pattern may be a combination of the data with the output from a pseudo-random binary sequence generator, or the data may be converted first into a bi-orthogonal pattern selected from among a plurality of such patterns and then randomized. The resulting modulated carrier frequency is then amplitude modulated based upon a human visual model so that the modulated carrier frequency amplitude is greater in areas of high image complexity and lower in areas of low image complexity. The subliminal data represented by the modulated carrier frequency is added to the video signal. At a decoder the data is recovered using appropriate correlation techniques by first filtering the video signal to increase the subliminal data component relative to the image component, then demodulating the data component to produce the data pattern, and then correlating the data pattern with the possible data patterns to identify the particular pattern and associated data word.
The objects, advantages and other novel features of the present inventi
Baker Daniel G.
Zink Scott E.
Desir Jean W.
Gray Francis I.
Peng John
Tektronix Inc.
LandOfFree
Transparent embedment of data in a video signal does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Transparent embedment of data in a video signal, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Transparent embedment of data in a video signal will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2452854