Electrical computers and digital processing systems: support – Multiple computer communication using cryptography – Particular communication authentication technique
Reexamination Certificate
1998-06-05
2001-12-18
Hayes, Gail (Department: 2766)
Electrical computers and digital processing systems: support
Multiple computer communication using cryptography
Particular communication authentication technique
C380S201000, C380S217000, C382S100000, C382S232000
Reexamination Certificate
active
06332194
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates generally to preventing unlawful copying of audio, video and other media that can be digitized and, more particularly, to a method for inserting a watermark into digitized data.
2. Prior Art
Methods are known to insert a watermark into an image or audio or video data. One purpose of such a watermark is for copy generation management, e.g. Allowing a single generation of copies to be made from a master, but no subsequent copies to be made from the first generation copies.
In such an application, there are three states that must be represented “copy never”, “copy once” and “copy no more”. The “copy never” state is intended to prohibit any copying of the content. The “copy once” state is intended to permit a single generation of copying. As such, this one generation copy must have its state changed from the original “copy once” to the “copy no more” state.
The copy generation management can be implemented in a variety of ways. One solution is to use two different watermarks to encode the states “copy never” and “copy once” in the original material. Then, when a copy is made, an additional watermark is inserted that indicates the state “copy no more”. Several variations of this method are possible. First, the “copy once” watermark may be removed prior to the insertion of the “copy no more” watermark. In practice, this is probably unnecessary since more than one watermark can co-exist in the data. Also, if the watermark is easily removable, then it may be straightforward to circumvent the copy generation management system by simply removing the “copy once” watermark from the content. Content with no watermark is assumed to be freely Copiague.
Second, the “copy once” state may be represented by the existence of two watermarks. One watermark would be difficult to remove and, when detected in isolation, would represent the “copy no more” state. The second watermark would be easy to remove and would only have meaning when detected in conjunction with the “copy no more” watermark, detection of both watermarks would represent the “copy once” state. An advantage of this approach is that, if carefully designed, the fragile watermark would be automatically removed by the existing installed base of VHS tape recorders. Note, however, that the fragile watermark must still survive signal processing that is common to video post production and transmission, e.g. low pass filtering, color correction and MPEG-2 compression.
Watermarks are typically inserted as noise in the digital signal during the production stage. The insertion of the watermark can be accomplished by many different procedures known in the art. The watermark, although inserted as noise, must be invisible to human perception or the image quality will suffer. Thus watermarks inserted at the production stage are image dependent and require great computational effort to ensure that the watermark will be invisible but robust enough so that it will survive post signal processing.
Once inserted, it is necessary to detect the presence of the watermark at a subsequent time. A typical method for determining the presence of a watermark in suspect data is to compare some characteristic of the data or signal derived from the data characteristics to a target watermark pattern. The watermark is determined to be present if this comparison indicates sufficient similarity.
A typical method of the prior art for watermarking data is illustrated in FIG.
1
.
FIG. 1
shows three similar data sets and, above each, the signal that would be derived from the data characteristics during detection. To the original data
1
a watermark pattern
2
is added. The sum of these two signals is the watermarked data
3
. During the detection process a signal derived from data characteristics
6
is extracted from the watermarked data
3
and compared to a target watermark pattern
5
. The watermarked data
3
is the sum of the original data
1
and the watermark pattern
2
. This may be a predetermined linear or non-linear combination. The signal derived from the data characteristics
6
, in this example is a linear (or perhaps non-linear) combination of a signal derived from the original data
4
and the target watermark pattern
5
. In general, the characteristics of the watermark and those of the original data will interact and thus the signal derived from the characteristics of the watermarked data
6
and that derived from the characteristics of the watermark
5
will not exactly match. The similarity of the signal derived from the characteristics of the watermarked data
6
and the signal derived from the characteristics of the watermark
5
indicates the likelihood that the data contains the watermark
2
. Note that in this illustration, the signal derived from the characteristics of the watermarked data
6
is similar to the signal derived from the characteristics of the watermark
5
suggesting the watermark
2
is present in the watermarked data
3
, but the signal derived from the original data
4
and the signal derived from the characteristics of the watermarked data
6
are dissimilar, indicating that the original data
1
does not yet contain the watermark
2
.
Note that in
FIG. 1
, it is desirable for the watermarking process to generate watermarked data
3
such that the extracted signal
6
is maximally similar to the target watermark
5
. In many cases this similarity is increased as the relative amplitude or strength of the watermark pattern
2
, and thus the signal extracted from the data characteristics
5
, is increased. The drawback to increasing the watermark strength is that this also tends to decrease the similarity between the watermarked and original data
3
,
1
, respectively. Generally, in the case where the data represents audio, video, or imagery, this similarity is judged by the human visual and auditory systems. Thus, increasing the watermark strength often leads to a loss of audio, video, or image quality.
A typical approach of the prior art to addressing these apparently contradictory concerns is to modify the watermark pattern such that when inserted at high enough strength for detection it introduces minimal perceptual degradation.
In view of the prior art, there is a need for a method for inserting a watermark into digitized data for copy generation management which can be done at the recorder stage and thus requires minimal computational effort and associated costs. There is also a need for a method which inserts a watermark of a significantly lower strength then previously used in the prior art that can be detected with high likelihood.
SUMMARY OF THE INVENTION
Therefore it is an object of the present invention to provide a method for data preparation and watermark insertion which requires a minimal computational effort at the recorder stage.
It is a further object of the present invention to provide a method for data preparation such that subsequent watermark insertion is inexpensive as compared to the methods of the prior art.
It is yet a further object of the present invention to provide a method for data preparation and watermark insertion which results in a watermark of a significantly lower strength, then previously done in the prior art, which can be detected with high likelihood.
It is still yet a further object of the present invention to provide a method for data preparation and watermark insertion for copy generation management which addresses all of the above objectives.
The method of the present invention differs from the typical approach in that not only is the watermark pattern carefully chosen to satisfy the tradeoff between high likelihood of detection and high perceptual similarity between watermarked and original data, but the method also modifies the characteristics and signals derived from the characteristics of the original data to reduce the interaction between the data characteristics and the watermark characteristics. This allows for a watermark of a significantly lower strength to be detected with high likelihood
Bloom Jeffrey A.
Cox Ingemar J.
Miller Matthew L.
Hayes Gail
Scully Scott Murphy & Presser
Signafy, Inc.
Song Ho S.
LandOfFree
Method for data preparation and watermark insertion 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 for data preparation and watermark insertion, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for data preparation and watermark insertion will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2566284