MXPA06009116A - Watermark detection by correlation shape analysis - Google Patents

Abstract

A watermark detector (100) detects a watermark in an information signal. The information signal is correlated with an expected watermark (Wi) for each of a plurality of relative positions of the information signal with respect to the watermark to derive a set of correlation results (64). The correlation results (64) are analysed to identify a cluster of correlation results which exceed a threshold value, the cluster representing a possible correlation peak. Where multiple clusters are identified, the most likely cluster is selected for further processing while other results are discarded. The cluster of results can identify a correlation peak which has become smeared due to lossy processing during distribution of the information signal.

Description

DETECTION OF WATER MARKS BY MEANS OF CORRELATION FORMAT ANALYSIS FIELD OF THE INVENTION This invention relates to detecting a watermark in an information signal. Background of the Invention Watermarking is a technique in which a mark of a certain type is added to an information signal. The information signal to which the watermark is added may represent a data file, a still image, video, audio or any other type of media content. The tag is inserted into the information signal before the information signal is distributed. The label is normally added in a manner that is imperceptible under normal conditions, such that it does not degrade the information signal, for example a watermark added to an audio file should not be audible under normal listening conditions. However, the watermark must be robust enough to remain detectable even after the information signal has undergone normal processes during transmission, such as coding or compression, modulation and so on. Many watermarking schemes employ correlation as a detection technique, with a low signal REF: 173755 test being correlated with a signal that contains a known watermark. In these systems, the presence of a watermark is indicated by one or more peaks in the correlation results. The document "A Video atermaking System for Broadcast Monitoring", Ton Kalker et al. , Proceedings of the SPIE, Bellingham, Virginia vol. 3657, January 25, 1999, p. 103-112, describes a scheme for detecting the presence of a watermark in transmitted video content. In most applications watermarked content will undergo several processing operations between the point at which "a watermark is inserted into the content and the point at which the presence of the watermark is detected. Common example of content processing is lossy compression, such as MPEG encoding Typically, the processing effects will reduce the correlation peaks that would normally be expected to occur during the watermark detection process. The present invention seeks to provide an improved way to detect a watermark in an information signal.As a result, a first aspect of the invention for detecting watermarks based on finding correlation peaks is considerably reduced. present invention provides a method for detecting a mark of water in an information signal, comprising: deriving a set of correlation results by correlating the information signal with a watermark for each of a plurality of relative positions of the information signal with respect to the watermark; analyzing the set of correlation results to identify a group of correlation results that exceeds a predetermined threshold value, the group represents a possible correlation peak. It has been found that the processing experienced by many information signals can have the effect of damaging a correlation peak when an attempt is made to detect the watermark by correlation. By identifying appropriately sized correlation result groups, it is possible to identify content with watermarks, even when processing or other attacks have degraded the quality of the watermark, reducing the height of the correlation peak below the threshold normally used for the detection. This improves the performance of the watermark detector and the extraction of the payload from the watermark. The ability to detect watermarks that are only weakly present in an article of media content also provides the option to allow the watermark to be more loosely inserted into the content, thus reducing its visibility under inspection by fraudulent parties potentials, or reducing their perceptibility under normal observation conditions. Preferably, if the step of analyzing the correlation result set identifies a plurality of groups of correlation results, the method further comprises processing the groups to identify the group that is most likely to represent the true correlation peak. This processing can be restricted to the correlation result groups, rather than to the complete set of correlation results. This can considerably reduce the amount of calculations required, leading to faster analyzes and simpler (and cheaper) detector requirements. The group of correlation results, and their values, provides information about the shape of the correlation peak, which can be used to further improve the performance of the watermark detector. The shape of the peak can be better understood by viewing the correlation results in the form of a graph, with the correlation value being plotted as height over a baseline of the graph. The functionality described here can be implemented in software, hardware or a combination thereof. Accordingly, another aspect of the invention provides software to carry out the method. It will be appreciated that the software can be installed in the host device at any point during the life of the equipment. The software can be. stored in an electronic memory device, hard disk, optical disk- or other machine-readable storage medium. The software can be supplied as a computer program product in a machine-readable carrier or can be downloaded directly to the device via a network connection. Additional aspects of the invention provide a watermark detector for carrying out any of the method steps and an apparatus for displaying an information signal that responds to the output of the watermark detector. Although the described embodiment refers to processing of an image or video signal, it will be appreciated that the information signal may be data representing audio or any other type of media content. BRIEF DESCRIPTION OF THE DRAWINGS The embodiments of the present invention will now be described, by way of example only, with reference to the appended figures, in which: Figure 1 shows a known way of inserting a watermark into a content article . Figure 2 shows an arrangement for detecting the presence of a watermark in a content article. Figures 3 and 4 show tables of results of correlation to be used in the detection method. Figure 5 shows a set of correlation result data in graphical form to show the shape of the peak and Figure 6 shows an apparatus for presenting content incorporating the watermark detection apparatus. Detailed Description of the Invention As a background, and to understand the invention, a process for inserting a watermark will be briefly described, with reference to Figure 1. A watermark pattern w (K) is constructed using one or more basic watermark patterns w. When a data payload is to be carried by the watermark, a number of basic watermark patterns are used. The watermark pattern w (K) is selected according to the payload - a multi-bit K code - to be inserted. The code is represented by selecting a number of the basic patterns w and by decentering them from each other by a particular distance and direction. The watermark pattern w (K) combined represents a noise pattern that can be added to the content. The watermark pattern w (K) has a size of M x M bits and is typically much smaller than the content article. Consequently, the pattern M x M is repeated (grid) 14 in a larger pattern that matches the format of the content data. In the case of an image, the pattern w (K) is squared 14 in such a way that it equals the size of the image with which it will be combined A content signal is received and stored in a volatile memory 16. A measure of local activity? (X) in the content signal is derived at each pixel position. This provides a measure of the additive noise visibility and is used to scale the watermark pattern W (K). This prevents the watermark from being noticeable in the content, such as equal brightness areas in an image. A total scaling factor s is applied to the watermark in the multiplier 22 and this determines the total resistance of the watermark. The choice of s is a compromise between the degree of robustness that is required and the requirement of how perceptible the watermark should be. Finally, the watermark signal (K) is added 24 to the content signal. The resulting signal, with the watermark inserted in it, will then be subjected to several processing steps as part of the normal distribution of that content. Figure 2 shows a schematic diagram of a watermark detector 100. The watermark detector receives content that may have watermarks. In the following description the content is assumed to be images or video content. The detection of watermarks can be carried out for individual tables or for groups of tables. The accumulated tables are divided into blocks of size M x M (for example, M = 128) and then they are folded into a volatile memory of size M x M. These initial stages are shown as block 50. The data in the volatile memory is then subjected to a Fourier Transformation Rapid 52. The next stage in the detection process determines the presence of watermarks in the data contained in the volatile memory. To detect whether the volatile memory includes a particular watermark pattern W or not, the contents of the volatile memory and the expected watermark pattern are correlated. Since the content data may include several watermark patterns, a number of parallel branches 60, "61, 62, each carrying out the correlation with one of the basic watermark patterns WO, Wl, are shown. W2 The correlation values for all possible displacement vectors of a basic pattern Wi are calculated simultaneously The basic watermark pattern Wi (i = 0, 1, 2) is subjected to a Fast Fourier Transform (FFT) before its correlation with the data signal The set of correlation values is then subjected to a Reverse Fast Fourier transformation 63. Complete details of the correlation operation are described in US 6,505,223 Bl. The Fourier coefficients used in the correlation They are complex numbers, with a real part and an imaginary part, representing a magnitude and a phase.It has been found that the reliability of the detector is significantly improved if the magnitude information is discarded and only the phase is considered. A magnitude normalization operation can be carried out after the multiplication by points and before the inverse Fourier Transformation 63. The operation of the normalization circuit comprises dividing by points each coefficient between its magnitude. The prior art is generally referred to as Match Only Filtering of Symmetric Phases (SPOMF). The set of correlation results from the "previous processing is stored in a volatile memory 64 and then analyzed by a group search operation. 65. Watermarked content is indicated by the presence of peaks in the correlation results data. It is highly probable that the peaks occur in pure Gaussian noise. The set of correlation results is examined to identify peaks that could be due to a watermark. The presence of a watermark can be indicated by a pronounced and isolated peak of significant height, although the more isolated peaks tend to represent spurious coincidences due to noise. A peak due to a watermark is more likely to be spotted or marked on several adjacent positions in the correlation results. The algorithm described below identifies potential watermark correlation peaks when searching for clusters of points of significant height that are closely spaced. The purpose is Find a group of points with an extremely low probability of occurrence. The grouping algorithm forms a number of groups of points, any of which may correspond to the actual correlation peak. The probabilities of these groups are compared, and the group with the lowest probability is assumed to be the desired correlation peak. The algorithm comprises the following steps: 1. Establish a threshold value and find all the points in the correlation data that are above this threshold value. All points that satisfy this criterion are stored in a list -ptsAboveThresh. A suggested threshold value is 3.3s (s = standard deviation of results in volatile memory 64) although this can be set to any preferred value. A preferred scale is 2.5 - 4s. If the threshold value is set too low under a large number of points, which do not correspond to the presence of a watermark, they will be stored in the list. Conversely, if the value is set too high there is a risk that the points corresponding to a valid, but spotted, peak will not be added to the list. 2. Find the point with the highest absolute value. 3. Form candidate groups, that is, groups of correlation points. Candidate groups are formed by collecting points that not only have a 'significant' value (a value greater than the threshold), but also locate very close to at least some other point of significant value. This is accomplished as follows: (i) The first point of the pts? BoveThresh list is removed and entered as the first p point of a new group; (ii) Look for pts? boveThresh points that are within a distance d from point p. All these points are removed from the ptsAboveThresh list and added to the group; (iii) The next point in the group is taken as the current p point. Step (ii) is repeated to add to the group all the points in ptsAboveThresh that are within the distance d of the new point p. '(iv) Step (iii) is repeated until ptsAboveThresh has been processed for all points in the group; (v) If the resulting group consists of a single point and that point is not equal to the highest peak found in the previous stage 2, then this group is discarded; (vi) Steps (i) to (v) are repeated until ptsAboveThresh is empty. At the end of this procedure, all the points originally entered in ptsAboveThresh in stage 1 above have been either: - assigned to a group that contains other points in the ptsAboveThresh list that are close to it, or - discarded, since they are not they have neighbors of similar height, and therefore are not part of a group.

A group is only allowed to understand a single point if that point has the highest absolute height of all points in the volatile correlation memory. This prevents a pronounced and non-stained correlation peak from being discarded, but prevents other isolated peaks, representing real noise, from being used. A final stage - detection of valid peaks 66 determines which of the result groups is most likely to represent the true correlation peak due to the presence of a watermark. There are several ways to achieve this. One technique, which is described in a co-pending patent application, compares the shape of the result set with stored data representing an expected peak shape. The comparison can be carried out by a cross-correlation technique. When there are several candidate groups, the comparison is carried out in each candidate group and the group that exhibits the closest match is selected as the one that represents the real correlation peak. Figures 3 and 4 show some exemplary sets of correlation data of the type that could be calculated by the detector. In the data set shown in Figure 3 the values vary between -3.8172 and 4.9190. It will be noted that watermarks can be inserted with negative amplitude. The highest value of 4.9190 is shown within box 130. Although this is below the typical detection threshold of 5, the highest value is surrounded by other correlation values of a similar value. This is indicative of a spike that has been spotted by processing during the distribution chain. Following the procedure described above, and establishing a threshold T of 3.3 and a distance of 1, it can be found that the correlation values within the ring 140 satisfy this criterion. It should be noted that the threshold is an absolute value and therefore the results - 3.8175 and -3.4377 are also included. When working through the process, the results of significant value are all located along one another. An isolated point, shown as point 142, is discarded during the process since it has no adjacent points on the threshold, and point 142 is not itself the highest point within the volatile memory. Looking at the data shown in Figure 4, the values vary between -3.7368 and 10.7652. Applying the same detection criteria, only one point 160 exceeds the threshold. The value of this point clearly exceeds the threshold and is thus considered to be a valid peak. From inspecting the adjacent values, it can be seen that this represents a pronounced correlation peak. Once a valid peak has been identified in one or more correlation data sets, the matching of the different data sets occurs to find a vector among the watermark patterns, that is, for identify the distance and direction in which the different patterns wO, wl, w2 are off center from each other. In a final step 75, the vectors identified in the previous step 70 are converted into a code, which represents the payload of the watermark. To illustrate what is intended by the form of a correlation peak, FIG. 5 shows a set of correlation results data plotted as a graph. In this example, a peak of -4.23 is displayed. If it is known that a content signal is likely to have a particular peak-correlation form, the threshold used in step 56 can be varied accordingly. For example, if it is known that the correlation peak will be high and pronounced, the threshold can be set high, whereas if it is known that the peak can be flattened, the threshold can be lowered in such a way that it does not prevent some correlation result Represent the real peak be excluded. Processing such as lossy compression, modulation and coding may flatten or otherwise distort the shape of the correlation peak. The inserted information represented as payload code K can identify, for example, the owner of copying rights or a description of the content. In the protection of DVD copies, allows the material to be labeled as "to be copied once",? To never copy ',? without restrictions', 'no more copies', etc. Figure 6 shows an apparatus for removing and displaying a content signal that is stored in a storage medium 200, such as an optical disk, memory device or hard disk. The content signal is removed by a content removal unit 201. The content signal 202 is applied to a processing unit 205, which decodes the data and validates it for display 211, 213. The content signal 202 also it is applied to a watermark detection unit 220 of the type previously described. The processing unit 205 is arranged in such a way that it is only allowed to process the content signal if a predetermined watermark is detected in the signal. A control signal 225 sent from the watermark detection unit 220 informs the processing unit 205 whether the processing of the content should be allowed or denied, or informs the processing unit 205 of any copying restriction associated with the content . Alternatively, the processing unit 205 may be arranged such that it is only allowed to process the content signal if a predetermined watermark is not detected in the signal. In the above description, a set of three watermarks has been considered. However, it will be appreciated that the technique can be applied to find a correlation peak in content data that carries only a single mark of water, or in content data that carry any number of different watermarks. In the above description, and with reference to the figures, there is described a watermark detector 100 which detects a watermark in an information signal. The information signal is correlated with an expected watermark Wi for each of a plurality of relative positions of the information signal with respect to the watermark to derive a set of correlation results 64. The correlation results 64 are analyze to identify a group of correlation results that exceed a threshold value, the group representing a possible correlation peak. When multiple groups are identified, the most likely group is selected for further processing while other results are discarded. The result group can identify a correlation peak that has been spotted due to loss processing during the distribution of the information signal. It is noted that in relation to this date, the best method known to the applicant to carry out the aforementioned invention, is that which is clear from the present description of the invention.

Claims (12)

1. CLAIMS Having described the invention as above, the content of the following claims is claimed as property: 1. A method for detecting a watermark in an information signal, characterized in that it comprises: deriving a set of correlation results by correlating the signal of information with a watermark for each of a plurality of relative positions of the information signal with respect to the watermark and analyzing the set of correlation results to identify a group of correlation results that exceeds a predetermined threshold value , the group represents a possible correlation peak.
2. 2. The method according to claim 1, characterized in that the step of analyzing the result set comprises determining all the correlation results in the set that exceed the threshold value, and then determining which of those correlation results are located within a predetermined distance from each other.
3. 3. The method according to claim 1 or 2, characterized in that, if the step of analyzing the set of correlation results identifies a result of Isolated correlation exceeding the threshold value, the method further comprises determining whether that isolated correlation result is the correlation result that has the highest value within the set of correlation results. . The method according to any of the preceding claims, characterized in that, if the step of analyzing the set of correlation results identifies a plurality of groups of correlation results, the method further comprises processing the groups to identify the group that is most likely that represents the peak of real correlation. 5. "The method according to claim 4, characterized in that the processing comprises comparing the shape of the group of correlation results with information in stored form and selecting the group with the best match to the information in stored form. according to claim 4 or 5, characterized in that all the groups, other than the one selected as being the most probable, are discarded 7. The method according to any of the preceding claims, characterized in that the threshold value is varied. 8. Software characterized in that it is to carry out the method of conformity with any of the previous claims. A watermark detector for detecting a watermark in an information signal, characterized in that it comprises: means for deriving a set of correlation results by correlating the information signal with a watermark for each of a plurality of relative positions of the information signal with respect to the "water mark" and means for analyzing the set of correlation results to identify a group of correlation results that exceeds a predetermined threshold value, the group represents a possible correlation peak 10. The watermark detector according to claim 9, characterized in that it further comprises means for carrying out any of the steps of the method according to claims 2-7. with claim 9 or 10, characterized in that the means for deriving a set of correlation results, and the means to analyze the set of correlation results comprise a processor that is arranged to execute software to carry out those functions. 12. Apparatus for presenting an information signal comprising means for disabling the operation of the apparatus depending on the presence of a valid watermark in the information signal, the apparatus is characterized in that it comprises the watermark detector according to any of claims 9-11.