JP2007074454A - Method and apparatus for detecting digital watermark - Google Patents

Abstract

【Task】
The prior art aiming at improving the detection accuracy of a digital watermark for video is characterized by determining whether or not the frame is to be detected according to the degree of remaining watermark of each frame. However, the prior art does not disclose a method for estimating the degree of remaining watermark, and is unclear as a means for improving detection accuracy. In addition, since whether or not a detection target is possible is determined in units of frames, it is insufficient as a technique for improving the detection accuracy of watermarks that remain unevenly in a frame.
[Solution]
The frame image is divided into a plurality of image areas, an error rate corresponding to each frame image is estimated, and a subset having the minimum error rate is set as a watermark detection target.
[Selection] Figure 4

Description

  The present invention relates to a digital watermark technique for digital data, particularly moving image data.

  From the viewpoint of copyright protection for digital data such as image data, digital watermark technology has attracted attention. The digital watermark technique is a technique for embedding predetermined information in digital data according to a predetermined rule so that the predetermined information cannot be extracted from the digital data without using at least the predetermined rule. For example, according to a predetermined rule, if information related to the purchaser of the video content is embedded in each frame image of the video content in an invisible form and illegally copied, it is illegally copied according to the predetermined rule. By extracting the embedded information from the stored content, the person who made the illegal copy is identified. Non-patent document 1 describes a digital watermark technique for moving images.

  One of the technical problems required for moving image digital watermarking is to improve image processing tolerance. This is because the embedded information can be detected correctly even after the image in which the information is embedded is processed and edited. For example, at the time of distribution of image / video content, after embedding information in the image with a digital watermark, video compression such as MPEG is often performed in order to reduce the data capacity of the package and the transmission amount of the network. In addition, in the secondary distribution, in addition to the above-described image processing and the like, a complicated geometric deformation intentionally performed by an unauthorized person in order to remove the digital watermark is assumed. Therefore, it is important that the embedded information can be accurately detected even after these processes are performed.

  In the process of embedding a moving image digital watermark, an image pattern representing information is superimposed on each frame constituting a moving image, and the image pattern is generally the same in each frame. Therefore, in order to improve image processing tolerance, the image noise associated with the image pattern is reduced by adding a plurality of frames in which the watermark is embedded when detecting the moving image digital watermark, and the image pattern (that is, the embedded information) is detected. A technique for improving accuracy is proposed in Non-Patent Document 2.

Specifically, the detection device described in Non-Patent Document 2 holds a stack for storing S frame images and accepts input of frame (or field) images at regular intervals. In the stack of the detection device, the frame images are arranged in descending order of the remaining degree of the watermark. When a new frame image in which the remaining degree of the watermark is higher than the lowest frame image in the stack, the stack is updated. Is done. Moreover, the said detection apparatus performs a detection process according to the following procedure for every fixed time.
(1) Watermark detection in the top frame image of the stack (2) Watermark detection by adding the top frame of the stack and the second frame image
(S) Watermark detection by adding the frame images from the top of the stack to the bottom (Sth). In the above procedure, when any watermark is detected, the process is stopped and the detection result is output.

I.Echizen, H.Yoshiura, T.Arai, H.Kimura and T.Takeuchi: General Quality Maintenance Module for Motion Picture Watermarking, IEEE Transactions on Consumer Electronics Vol.45, No.4, pp.1150-1158, NOVEMBER 1999 Sugihara Ryo: Reliability evaluation of digital watermarking of video, 2000 Symposium on Cryptography and Information Security, SCIS2000-N02, (2000).

  Non-Patent Document 2 does not disclose a method for calculating the degree of remaining watermark in a frame image, and is lacking in specific content.

  Further, in Non-Patent Document 2, in the above-described detection processing procedure, the detection processing is stopped and a detection result is output when any watermark is detected. Therefore, the detection result is not necessarily optimal from the viewpoint of reducing the error rate. However, it is not sufficient as means for improving detection accuracy.

  Further, the remaining degree of the watermark of the frame image is generally non-uniform in the frame, but the detection process of Non-Patent Document 2 is based on the premise of the process in units of frames, and the non-uniformity of the remaining degree of the watermark in the frame Did not consider. That is, the degree of remaining watermark is calculated in units of frames, the stack frame images are rearranged based on the strength, and addition processing is performed, so noise reduction in the frames is not considered, and from the viewpoint of the processing units Even if it sees, it was inadequate as a means of improving detection accuracy.

  According to the present invention, in the detection of a moving image digital watermark, a region having a high degree of remaining watermark is selected and added so that the detection accuracy of watermark information from the moving image content is optimized. Specifically, a subset having the smallest estimated error rate is selected from a plurality of target frame images and added together.

A desirable aspect of the present invention is a digital watermark information detection method for detecting information inserted as a digital watermark from image content.
A first step of dividing the image content into at least two or more image regions;
A second step of estimating a remaining degree of digital watermark information corresponding to each divided image area;
A third step of rearranging the corresponding image area or the statistical value set corresponding to the image area based on the estimated remaining degree;
A fourth step of sequentially reordering the rearranged image area or statistical value set in ascending or descending order and re-estimating the remaining degree of the digital watermark information from the added image area or the added statistical value set for each addition; ,
A fifth step of selecting a plurality of added image regions, or an image region having a minimum remaining degree from the added statistical value set, or a statistical value set;
And a sixth step of determining digital watermark information from the selected image region or statistical value set.

  Furthermore, in the digital watermark information detection method, the image content may be a video content composed of two or more frame images.

  Further, in the digital watermark information detection method, the statistical value set corresponding to the image area is configured with statistical values corresponding to each information bit constituting the digital watermark information, and each statistical value corresponds to each information bit. It may be expressed by a product of a random number and a pixel value on the image area.

  Furthermore, in the digital watermark information detection method, corresponding image regions or statistical value sets may be rearranged so that the remaining degree is in descending order or ascending order.

  According to the present invention, it is possible to improve the digital watermark image processing resistance without changing the moving image digital watermark embedding process.

  FIG. 1 is a diagram showing a schematic configuration of the present embodiment. As shown in FIG. 1, as a content distribution method, a method via a broadcast, a package, or a network is the mainstream, and the application form of this embodiment is a digital watermark embedding / control signal adding apparatus 100 and a digital watermark detection / control. A signal detection device 200 is included.

  The digital watermark embedding / control signal adding apparatus 100 is an apparatus that embeds digital watermark information in content and adds a control signal during content creation or content distribution, and may have a content editing and content distribution function. The digital watermark detection / control signal detection device 300 is a device that detects digital watermark information and a control signal added to the content at the time of content reproduction or content reception, and is a content reproduction and content reception function or digital watermark embedding.・ A control signal addition function may be provided.

  FIG. 2 is a diagram showing a schematic configuration of the digital watermark embedding / control signal adding apparatus 100 of the present embodiment. As shown in FIG. 2, the digital watermark embedding / encrypting apparatus 100 of this embodiment includes an information processing apparatus having a CPU 101, a ROM 102, an input device 103, an output device 104, a communication device 105, and a memory 106. Can be configured.

  The CPU 101 is a device that controls the overall operation of the digital watermark embedding / control signal adding device 100. The memory 106 is a storage device that stores various processing programs and data for controlling the operation of the entire digital watermark embedding / control signal adding apparatus 100. The input device 103 is a device that performs various inputs for inputting content. The output device 104 is a device that performs various outputs accompanying content output. The communication device 105 is various devices that communicate with and control other processing devices via a network such as the Internet or an intranet, or a dedicated interface such as SCSI, RS422, or IEEE1394.

  The digital watermark embedding / control signal adding apparatus 100 executes a content reading processing unit 108, a digital watermark embedding processing unit 109, a control signal adding processing unit 110, and a content output unit 111 by being executed by the CPU 101. A program 120 is stored in the memory 106.

  The content read processing unit 108 reads a part or the entire area of the content from the input device 103 or the storage area 107.

  The digital watermark embedding processing unit 109 is a case where the read content part or the entire region is a region to which control information for controlling the processing contents of the content in the digital watermark detection / control signal detection device 200 is added. , Recording / playback of content, storage in non-processed area, storage in processable area, display of accompanying data accompanying content, selection of content sending device, or bidirectional communication between sender and receiver Information such as control information, user ID and receiver ID, user level information conditions, and content source ID are embedded in the area as digital watermark information.

  The control signal addition processing unit 110 adds content to a region where a control signal for controlling the processing contents of the content in the digital watermark detection / control signal detection apparatus 200 is added to a specific portion such as a header or a footer in the content. Recording / playback, storage in non-processed area, storage in processable area, display of accompanying data accompanying content, selection of content sending device, or bidirectional communication between sender and receiver Control information, user ID and receiver ID, user level information conditions, and information such as the sender ID of the content are consistent with the control information embedded by the digital watermark embedding processing unit 109. However, it is added to the specific location of the content as a control signal.

  In this embodiment, the processing procedures of the digital watermark embedding processing unit 109 and the control signal addition processing unit 110 are the digital watermark embedding processing first and the control signal addition later. There is no need to process, and the control signal may be added first.

  The content output unit 111 transmits the content to which the control signal is added by the control signal addition processing unit 110 to the output device 203. In addition, the content output unit 111 may transmit content on the storage area 107 that has not been encrypted and / or watermark embedded (or one of them) to the output device 104.

  FIG. 3 is a diagram showing a schematic configuration of the digital watermark detection / control signal detection apparatus 200 of the present embodiment. As shown in FIG. 3, the digital watermark detection / decryption apparatus 200 according to the present embodiment includes an information processing apparatus having a CPU 201, a ROM 202, an input device 203, an output device 204, a communication device 205, and a memory 206. Can be configured.

  The CPU 201 is a device that controls the overall operation of the digital watermark detection / control signal detection device 200. The memory 206 is a storage device that stores various processing programs and data for controlling the operation of the entire digital watermark detection / control signal detection apparatus 200. The input device 203 is a device that performs various inputs for inputting content. The output device 204 is a device that performs various outputs accompanying content output. The communication device 205 is various devices that perform communication and control with other processing devices via a network such as the Internet or an intranet, or a dedicated interface such as SCSI, RS422, or IEEE1394.

  Further, the electronic watermark detection / control signal detection apparatus 200 is executed by the CPU 201, so that the content read processing unit 208, the control signal detection processing unit 209, the digital watermark detection processing unit 210, and the detection information consistency determination unit 211 are executed. The memory 206 has a program 220 for realizing the control processing unit 212.

  The content read processing unit 208 reads a part or the entire area of the content from the input device 203 or the storage area 207.

  The control signal detection processing unit 209 controls specific contents such as a header and a footer in the content read by the content reading processing unit 208 to be processed by the digital watermark detection / control signal detection device 200 of the content. Recording and playback of content, storage in a non-processed area, storage in a processable area, display of accompanying data associated with the content, selection of content sending device, or transmission when the control signal is added Specify control information such as control information for controlling two-way communication between the receiver and receiver, user ID and receiver ID, user level information conditions, and source ID of the content Detect from location.

  The digital watermark detection processing unit 210 performs detection processing of a digital watermark from a part or the entire area of the content, and is added with control information for controlling the processing content of the content in the digital watermark detection / control signal detection device 200. Content recording / playback, storage in a non-processed area, storage in a processable area, display of accompanying data associated with the content, selection of content transmission equipment, or transmission side and reception side Information such as control information for controlling the two-way communication between the user, the user ID and the receiving ID, the condition of the user level information, and the transmission source ID of the content are detected as digital watermark information.

  In this embodiment, the processing procedure of the control signal detection processing unit 209 and the digital watermark detection processing unit 310 is preceded by the control signal detection processing and after detection of control information by the digital watermark. It is not necessary to perform the processing in the processing procedure, and the control information may be detected by digital watermark first.

  The detected information consistency determination unit 211 determines the consistency between the control information detected from the content by the control signal detection processing unit 209 and the control information detected from the content by the digital watermark detection processing unit 210, and based on the determination result. The processing instruction to the control processing unit 212 is determined.

  As a processing instruction for the content, the control processing unit 212 reproduces the content, copies the content, stores the content in a non-processed area, stores it in a processable area, displays accompanying data accompanying the content, and selects a content transmission device. Alternatively, an instruction to execute bidirectional communication between the transmission side and the reception side is received from the detection information consistency determination unit 211, and the received instruction is executed for the content.

  In the present embodiment, each of the above programs is stored in a magnetic disk (not shown) provided in the information processing apparatus constituting the digital watermark embedding / control signal adding apparatus 100 or the digital watermark detection / control signal detecting apparatus 200, It is assumed that the program is loaded into the memory 106 or 206 and executed. Further, the program may be used by installing it in the information processing apparatus from a recording medium such as a CD-ROM that can be used by the information processing apparatus, or by accessing another information processing apparatus via a network. It may be installed and used. Alternatively, the program may be stored in a magnetic disk or the like and loaded into the memory 106 or 206 using a network that can be used by the information processing apparatus from an external server or a carrier wave that propagates over the network. Each processing unit is not necessarily realized by executing a program, but may be realized by hardware having a circuit configuration such as an FPGA.

  4 and 5 are diagrams showing a schematic configuration and a processing outline of the digital watermark detection processing unit 210 in the digital watermark detection / control signal detection apparatus 200 of the present embodiment.

  As shown in FIG. 4, the digital watermark detection processing unit 210 of the present embodiment includes an area dividing unit 41, a statistical value calculation unit 42, an error rate estimation unit 43, a sort processing unit 44, an addition control unit 45, The watermark information determination unit 46 is configured.

  As shown in FIG. 5, the region dividing unit 41 divides a predetermined number of frame images 40 into a predetermined number (for example, N) of image regions 51.

  The statistical value calculation unit 42 sets, for each image region divided by the region dividing unit 41, a statistical value set V = {V (k) | 1 ≦ k ≦ corresponding to the embedded information bit Bk (1 ≦ k ≦ M). M} is calculated. Examples of calculating each statistical value V (k) include V (k) = ΣP (k, i) × M (k, i). Here, Σ means that the sum is taken for the subscript i, and M (k, i) represents a pseudo-random number sequence independent of the subscripts k and i.

  The error rate estimation unit 43 estimates the degree of error of the embedded information bits from the statistical value set V calculated from each image region. As a specific estimation method, for example, there is a method described in JP-A-2005-033818. This estimation method is based on a probability distribution curve determined by the distribution of at least two statistics values among a plurality of statistics values constituting a statistics value set, and the bit data when the threshold value used for the determination of the bit values of the statistics values is T This is a method for obtaining the bit value determination error rate of the.

  The sort processing unit 44 sorts the N statistical value sets calculated from the N image regions based on the error rate estimated from each statistical value set.

  The addition control unit 45 performs addition processing as shown in FIG. 5 for the statistical values of each sorted statistical value set. That is, added statistical value set 1 = statistical value set 1, added statistical value set 2 = statistical value set 1 + statistical value set 2, added statistical value set 3 = statistical value set 1 + statistical value set 2 + statistical value set 3,. , The added statistical value set N = statistical value set 1+... + The statistical value set N is calculated by processing. Here, the operation “+” means that an addition operation is performed on each element (statistical value) of the set.

  Next, the addition control unit 45 estimates an error rate of the embedded information bits for each generated set of statistical values. The estimation method follows the method described in Patent Document 1.

  Next, as shown in FIG. 5, the addition control unit 45 selects an addition statistical value set that minimizes the error rate from the N addition statistical value sets.

The watermark information determination unit 46 determines the watermark information using the threshold value T used for bit value determination from the added statistical value set selected by the addition control unit 45, and outputs watermark information 47. As an example of bit value determination,
Bk = 1 if V (k) ≧ T, Bk = 0 if V (k) <T (1 ≦ k ≦ M) can be considered.

  FIG. 6 is a flowchart showing the processing procedure of the digital watermark detection processing unit 210.

  First, in step S600, the area dividing unit 41 reads one or more predetermined number of frame images.

  In step S601, the area dividing unit 41 divides the read predetermined number of frame images into a predetermined number (N) of image areas.

  Next, in step S602, the statistical value calculation unit 42 calculates a statistical value set for each divided image region, and obtains N statistical value sets.

  In step S603, the error rate estimation unit 43 estimates an error rate of information bits from each statistical value set, and obtains N error rates corresponding to the N statistical value sets.

  In step S604, the sort processing unit 44 sorts the N statistical value sets based on the error rate estimated from each statistical value set.

  In step S605, the addition control unit 45 performs the above-described statistical value addition process on the statistical values of each sorted statistical value set to obtain N additional statistical value sets.

  Next, in step S606, the addition control unit 45 re-estimates the error rate of information bits from each addition statistical value set, and obtains N error rates corresponding to the N additional statistical value sets.

  Next, in step S607, the addition control unit 45 selects an addition statistical value set that minimizes the error rate from the N addition statistical value sets.

  In step S <b> 608, the watermark information determination unit 46 determines the watermark information using the threshold value T used for bit value determination from the added statistical value set selected by the addition control unit 45, and outputs the watermark information 47.

  Next, in step S609, a control unit such as a CPU that controls the digital watermark detection processing unit 210 determines whether the digital watermark detection processing unit 210 has completed reading a predetermined number of frame images, and the reading processing has been completed. If there is, the process of the digital watermark detection processing unit 210 is terminated. If the reading process is not completed, the process returns to step S600, and reading of a predetermined number of frame images is started.

  In the present embodiment, video content is targeted as processing target content, but is not necessarily limited to this, and still image and music content can also be targeted.

  When still image content is targeted, in this embodiment, the image area is obtained by dividing the still image into a plurality of image areas, and the error rate of the watermark information in each image area is estimated. It is possible to determine whether or not an area is a watermark detection target.

  In addition, when music content is targeted, in this embodiment, a plurality of content areas are obtained by time-sharing the music content, and the error rate of the watermark information in each area is estimated, so that the area is It is possible to determine whether or not to make a watermark detection target.

  As described above, in the present embodiment, in the detection of the moving image digital watermark, an area having a high degree of remaining watermark is selected and added so that the detection accuracy of the watermark information from the moving image content is optimized. Specifically, a subset having the smallest estimated error rate is selected from a plurality of target frame images and added together. According to the present embodiment, it is possible to improve the digital watermark image processing resistance without changing the moving image digital watermark embedding process.

It is a figure which shows schematic structure of the system to which this embodiment is applied. It is a figure which shows schematic structure of the digital watermark embedding and control information addition apparatus of this embodiment. It is a figure which shows schematic structure of the digital watermark detection and control information detection apparatus of this embodiment. It is a figure which shows schematic structure of the digital watermark detection process part of this embodiment. It is a figure which shows the processing content of the digital watermark detection process part of this embodiment. It is a flowchart which shows an example of the process sequence of the digital watermark detection process part of this embodiment.

Explanation of symbols

40: frame image, 41: region dividing unit, 42: statistical value calculating unit, 43: error rate estimating unit, 44: sort processing unit, 45: addition control unit, 46: watermark information determining unit, 47: watermark information.

Claims (4)

A method of detecting digital watermark information for detecting information inserted as a digital watermark from image content,
Dividing the image content into at least two or more image regions;
A second step of estimating a remaining degree of the digital watermark information corresponding to each of the divided image regions;
A third step of rearranging a corresponding image area or a statistical value set corresponding to the image area based on the estimated degree of the remaining degree;
The rearranged image area or the statistical value set is sequentially added in ascending or descending order, and the remaining degree of the digital watermark information is re-estimated from the added image area or the added statistical value set for each addition. Steps,
A fifth step of selecting the plurality of added image regions, or an image region having the smallest remaining degree from the added statistical value set, or a statistical value set;
And a sixth step of determining the digital watermark information from the selected image region or statistical value set. The digital watermark information extracting method according to claim 1,
The method of detecting digital watermark information, wherein the image content is video content composed of two or more frame images. The method for extracting digital watermark information according to claim 1 or 2,
The statistical value set corresponding to the image area is composed of statistical values corresponding to information bits constituting the digital watermark information,
Each statistical value is expressed by a product of a random number corresponding to each information bit and a pixel value on an image area. A method for extracting digital watermark information according to any one of claims 1 to 3,
The third step is a method of detecting digital watermark information, wherein the corresponding image regions or statistical value sets are rearranged so that the remaining degree is in descending order or ascending order.

Images