Image analysis – Applications
Reexamination Certificate
1999-03-02
2003-04-08
Boudreau, Leo (Department: 2621)
Image analysis
Applications
C713S176000
Reexamination Certificate
active
06546113
ABSTRACT:
TECHNICAL FIELD
The present invention relates generally to steganography and, in particular, to techniques for embedding data in a video signal and for detecting the presence of such embedded data in a video signal.
BACKGROUND OF THE INVENTION
Steganographic techniques, as known in the art, generally attempt to hide one piece of information within another. Within the broad field of steganography, the ability to embed ancillary data in analog or digital data representing video or graphics has a number of uses. Such systems ideally hide authentication or other data in a data stream or store in such a way as to allow the data stream or store (for example a graphics file on a floppy disk or a video-tape) to function normally and without indicating the presence or absence of the embedded data. Toward the end of aiding in the identification of copyright infringement or source identification, technology that allows authentication data to be carried by the data itself is useful. Employing a video image or stream of images to serve as a “carrier” for embedded data is also useful for annotating and indexing the carrier data itself. An example, called digital watermarking, is a technique that modifies data representing a video image or stream of images in such a way that the modifications are essentially undetectable.
Various methods have been used in the past to “hide” digital information in a video signal. In many cases, the added data is associated with the video, but is not part of the video image itself For example, in U.S. Pat. No. 5,737,026 issued Lu et. al., the added data is encoded in an overscan region of the video signal, and therefore can be removed without affecting the video images. Such systems are clearly unsuitable for copyright protection where the added data must be substantially indelible. In other cases, the watermark data is encoded into the video in such a simple manner that straightforward methods may be used to detect and remove the data thereby destroying the watermark. For example, the “fingerprint” apparatus of U.S. Pat. No. 5,668,603 issued to Copeland et al. communicates each bit by raising or lowering the video set-up by 0.5 IRE for a complete field.
In U.S. Pat. No. 3,984,624 issued to Waggener et al., data is added to the video signal in randomized positions, making the data difficult to remove. However, no attempt is made in this case to hide the video data from the viewer, and the digital signature therefore impairs the video to some degree. A method for hiding data in a video data stream is disclosed in U.S. Pat. No. 4,969,041 issued to O'Grady, et al. A low-level waveform is added to the video signal during the active video portion of the video signal. The low-level waveform can be any unique waveform, such as a set of random noise waveforms that are unlikely to occur in a normal video signal, each of which represents a unique digital data word. The low-level waveform to be embedded in the video signal has levels significantly below the noise level of the video signal. The low-level waveform is detected by correlating the video signal with all of the unique waveforms, or with a desired one of the unique waveforms if a particular data word is sought. The video signal is multiplied by each waveform, or with the desired waveform, and the result is compared with a threshold value to determine a correlation coefficient for each waveform. This technique, however, is vulnerable to attack in that the video image can be modified in such a way as to make the signal uncorrelatable with the standard waveforms. For example, an adverse party could zoom the video images to distort the embedded signal to the point that it is no longer recognizable without destroying the image.
While the foregoing discussion of prior art techniques does not represent an exhaustive review of the state of the prior art, it does point out the shortcomings of current techniques. In particular, a need exists for a simple technique for embedding data in a video signal that minimizes the impact on the quality of the video signal, that is substantially impervious to attempts to remove or obscure the embedded data, and that provides for straightforward detection of the embedded data.
SUMMARY OF INVENTION
These needs and others are substantially met by the present invention. Generally, the present invention provides a technique for embedding data (a “watermark”) in a video signal. The watermark may be used to identify the source and/or the owner of the information in the video signal. The encoded video signal meets all broadcast and studio specifications, and has no visible impairments. Attempts to obscure or remove the embedded data by adding noise, coring, cropping or warping the image will damage the video to an unacceptable degree before the watermark is eliminated. The method is cascadable, and survives digital compression and/or conversion to analogue.
A low-level digital signal is added to a parameter defining the video signal. The watermark data consists of known two-dimensional codeword patterns that are added to portions of each video frame, and covering a large proportion of the frame. The portions can be, for example, alternate lines or alternate fields. The codeword patterns consist of relatively few bits per frame so that each bit of a codeword covers a significant portion of the frame. Therefore only low spatial frequencies are present in the codeword patterns. Limited randomization of the spatial position of the codeword pattern is also employed. By altering a polarity of a codeword pattern, binary data can be communicated on a per frame basis.
Detection of the codeword pattern is accomplished by a two step process. In the first step, a predictor is employed that continuously estimates the current value of the modified parameter. In one embodiment of the present invention, the predictor employs video samples taken from unmodified areas of the frame (areas to which no data has been added, e.g., alternate lines or fields) to provide the estimates. The estimated parameter values are then subtracted from the actual parameter values constituting the video signal, thereby producing a low-amplitude noise-like signal containing the embedded watermark data. In the second step, multiple correlators attempt to detect the presence of the codeword pattern in a range of spatial locations surrounding its nominal position in the image frame. The use of multiple detectors (spatially offset from the original codeword location) provides immunity to feasible attacks aimed at damaging the watermark, i.e., levels of zoom, cropping or warping of the signal (short of destroying the video signal) do not materially affect the delectability of the mark.
REFERENCES:
patent: 3796868 (1974-03-01), Kaul et al.
patent: 4267580 (1981-05-01), Bond et al.
patent: 5111292 (1992-05-01), Kuriacose et al.
patent: 5122875 (1992-06-01), Raychaudhuri
patent: 5613004 (1997-03-01), Cooperman et al.
patent: 5689587 (1997-11-01), Bender et al.
patent: 5768426 (1998-06-01), Rhoads
patent: 5850482 (1998-12-01), Meany et al.
patent: 6131161 (2000-10-01), Linnartz
patent: 6154571 (2000-11-01), Cox et al.
“Steganography”; www.indstate.edu/msattler/sci-tech/comp/privacy/topics/steganography.html.
“Steganography Info and Archive”; members.iquest.net/~mrmil/stego.html.
“Digital Watermarking and Image Authentication”; www.ctr.columbia.edu/—mars/papers/reports/water/doc.html.
“Look, It's Not There”; www.byte.com/art9701/sec/art1.htm.
“How to Embed a Watermark”; www.byte.com/art/9701/sec18/art1.htm.
“Steganography Overcomes Cryptography Restrictions”; www.byte.com/art/9701/sec18/art2.htm.
“Steganography”; patriot.net/~johnson/html
eil/stegdoc/stegdoc.html.
Lucas Keith
Van Rassel William
Banner & Witcoff , Ltd.
Boudreau Leo
Leitch Technology International Inc.
Tabatabai Abolfazl
LandOfFree
Method and apparatus for video watermarking does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for video watermarking, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for video watermarking will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3108327