Image analysis – Applications
Reexamination Certificate
2000-06-19
2003-12-16
Mehta, Bhavesh M. (Department: 2625)
Image analysis
Applications
C382S232000, C713S176000, C380S054000
Reexamination Certificate
active
06665420
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This application relates to the field of digital multimedia and more particularly to the field of watermarking digital video content.
2. Description of Related Art
The secure transmission of digital multimedia information is an important concern for multimedia content owners, particularly in distribution channels having public access, such as the Internet or wireless networks. One form of protection has been provided, traditionally, through encryption. However, for widespread distribution of the information, key-management for the encryption is cumbersome. Furthermore, encryption provides incomplete security since, after decryption, the original digital content may be readily reproduced and distributed.
Another form of protection is provided by a message authentication code (“MAC”) that is attached to digital content. For example, origination information can appear within a message field appended to digital content. However, this type of add-on authentication is easily identified and removed. In addition, if the communication channel is lossy, conventional cryptographic methods can be inadequate. For example, a single bit error can cause the received signal to fail an authentication check.
A more resilient MAC system is provided by digital watermarking. Digital watermarking is a technique for hiding an identification of origin in a digital media stream. A watermark is actually embedded in the media stream, so that removal of the watermark may destroy or visibly alter the underlying content. The watermark may also be inserted into the original data in a manner that is imperceptible to the listener/viewer. When such watermarked digital content is distributed on-line, or recorded on a disk, the origination data travels with it, and allows the originator to demonstrate the source of the content. Digital watermarking also identifies tampering, since any manipulation of the content will result in an invalid watermark.
There are several characteristics that define useful watermarks. One characteristic is that they be survivable to compression and common signal distortions that may occur during transit. If the watermark survives, it is difficult for the recipient to remove the watermark without destroying the content that it protects. Such a feature is important in protecting copyright rights. Another characteristic is that the watermark describe the content. For example, watermarking can be used to authenticate and tamper-proof multimedia content. For such a use, the watermark must still exhibit survivability (i.e., be robust to compression, common signal distortions, and the effects of error-prone transmission channels), but must also indicate whether any modifications were made to the content itself. Accordingly, the watermark should describe the content and the salient features of the video information. Known approaches for constructing video authentication MACs involve using edge maps or image histograms of the video frames.
Regarding the survivability characteristic of watermarks, for ownership-authentication applications such as for copyright protection, the list of video signal distortions that the MAC must survive is quite extensive, because the content owner is interested in claiming ownership regardless of the manipulation done to the original content. For example, such MACs must be robust to (i.e., survive) cropping, rotating, dithering, and recompression and the like. The list of acceptable distortions for authentication/tamper-proofing purposes is more limited, because the authentication MAC must still be able to identify tampering. Distortions caused by recompression or minor channel errors do not indicate tampering, while cropping, rotating and the like do. During recompression, the essence of the video information is maintained, but minor fluctuations in individual coefficient values, though perhaps not perceptible, can change the value of the MAC. For authentication, determining what distortions are acceptable and which constitute tampering is difficult.
There is therefore a need for a system and method for generating a message authentication code (MAC) that minimizes or eliminates one or more of the problems set forth above.
SUMMARY OF THE INVENTION
One object of the present invention is to provide a message authentication code that is robust to predefined signal distortions (e.g., at a level expected due to normal recompression, etc.).
Another object of the present invention is to provide a message authentication code that compactly describes digital video content in a manner that permits tamper-proofing and authentication of origin.
It is yet another object of the present invention to provide a message authentication code that permits identification of particular areas within an image that have been subjected to tampering.
To achieve these and other objects, a system and method of generating a message authentication code (MAC) is provided. One advantage of the present invention is that it provides a MAC having a built-in tolerance to expected signal distortions. In particular, the invention provides for an error tolerance buffer that is based on an amount of acceptable distortion that the video information is expected to incur during recompression, etc. Minor modifications in coefficients associated with the content that fall within the accepted error tolerance range will not cause a change in the MAC. Modifications above the given threshold, whether due to excessive recompression, or actual tampering, will alter the MAC and will indicate tampering in a receiver.
In accordance with the present invention, a method for generating a MAC associated with an image includes four basic steps. The first step involves receiving a plurality of blocks of image coefficient data, each coefficient having an original value in a range of values. The second step involves defining an error tolerance buffer by dividing the range of values into first and second regions and a buffer region therebetween. The first and second regions have allowed coefficient values while the buffer region contains disallowed coefficient values. The next step involves mapping an original value of at least one coefficient from each block to a respective modified value falling within one of the first and second regions, but not the buffer region (disallowed coefficients). The last basic step involves generating the MAC according to predetermined strategy as a function of the most significant bit (MSB) of the image coefficient values.
In a preferred embodiment, a DC coefficient from each block is mapped. The MSBs of low frequency (e.g., DC) coefficients are most likely to survive minor distortions such as recompression. Preferably, the predetermined strategy comprises any algorithm for producing a MAC that uses the MSBs of the DC coefficients. However, MACs based on the MSBs of DC coefficients are vulnerable, at least in one respect, to being lost, even though the essence of the low frequency information is maintained during recompression or the like. For example, for coefficients that range from 0 to 255, the coefficients 128 and 127 are close in value even though their bit representations differ in every bit. A small change in the value of a coefficient (e.g., from 127•128, due to recompression) could result in the MSB of a coefficient changing (e.g., MSB=0•MSB=1). The error tolerance buffer specifies disallowed coefficient values—ones that are vulnerable to minor distortions that would change their MSB and thus alter the MAC. All coefficients are mapped, but any coefficients in the buffer region are moved away to either the first region or the second region. After mapping, the modified coefficients all have at least a minimum predefined amount of “distance” to an MSB crossover threshold. For example, for an 8-bit coefficient, the crossover occurs between 127 and 128. For a 10-bit coefficient, between 511 and 512. So long as any change to the image coefficients is less than the minimum “distance”, the MAC will remain unchanged—o
Arce Gonzalo
Basch Evert
Lewis Arianne M.
Xie Liehua
Carter Aaron
Mehta Bhavesh M.
Suchyta Leonard Charles
Verizon Laboratories Inc.
Weixel James K.
LandOfFree
Message authentication code with improved error tolerance does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Message authentication code with improved error tolerance, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Message authentication code with improved error tolerance will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3122609