JP2006254100A - Detecting method and device for water mark information - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a device capable of efficiently detecting highly accurate electronic watermark information from images or video streams applied with data compression such as MPEG. <P>SOLUTION: An MPEG stream 101 from an input unit 6 is input into a quantize scale magnitude determining unit 1 for determining the magnitude of quantize scale and outputting a QS magnitude determination signal indicative of whether electronic watermark should be detected or not and an MPEE decoding portion 2 for decoding an inputted MPEG signal. The quantize scale magnitude determining unit 1 controls a switch 3 so as to supply a frame of smaller QS of the inputted MPEG signal to an electronic watermark extracting portion 4. Since the electronic watermark extracting portion 4 can use a frame of smaller QS for electronic watermark extraction, a highly accurate electronic watermark information can be detected. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a method and apparatus for detecting digital watermark information, and in particular, information that is difficult to see such as a digital watermark embedded in an image for the purpose of protecting the copyright of the image and preventing falsification of the image. The present invention relates to a detection method and apparatus for digital watermark information to be output.

  Compared to analog images, digital images that handle images as digital data can be easily copied without image quality degradation by a computer, etc., and can be sent via a communication line, etc. In addition, the image quality is not deteriorated. Due to such characteristics, a digital image has a great damage to a copyright owner when the image is illegally copied. As a conventional technique related to a method for preventing unauthorized copying of a digital image, there is a technique called digital watermark. For example, a technique described in Patent Document 1 is known.

  There are two types of digital watermarks: a non-visible type that embeds watermark information such as copyright information and user information in an invisible image, and a watermark image such as a logo of a company that owns the copyright on the image. There is a visible type that forms in a visible form.

  An invisible digital watermark cannot be recognized as having watermark information embedded at first glance, but is embedded by processing the image according to a specific method. Watermark information data such as copyright information can be extracted. The method according to the prior art described in Patent Document 1 described above is that watermark data such as copyright information is embedded in the entire image in the form of noise, and watermark data is taken out by statistically processing it. is there.

  Since this method is based on statistical processing, if the image data embedded with a watermark deteriorates for some reason, for example, if the entire image has noise, the watermark data that was originally embedded May not be extracted completely, or it may not be possible to guarantee that the extracted data is sufficiently reliable. In this case, for example, even though the copyright information is embedded as watermark data, sufficient protection of the copyright cannot be performed.

  One cause of image data degradation is compression processing on image data. A method called MPEG (Moving Picture Experts Group) is known as one of the compression methods for moving images widely used at present. MPEG is capable of efficiently compressing each frame of a moving image according to a plurality of compression methods. The compression methods used here include an intra-frame coding scheme and an inter-frame coding scheme. Further, the inter-frame coding scheme includes a forward predictive coding scheme in which the immediately preceding frame is a predicted frame, There is a bidirectional predictive encoding method in which the immediately preceding and immediately following frames are used as predicted frames. In MPEG, an image based on an intra-frame coding scheme is called an I picture, an image based on a forward predictive coding scheme is called a P picture, and an image based on a bidirectional predictive coding scheme is called a B picture.

  As a conventional technique that can solve the problem that the embedded watermark data cannot be correctly extracted due to the deterioration of the image data caused by the image compression process, for example, a technique described in Patent Document 2 is known. Are known. The prior art described in Patent Document 2 assumes that only some still images of moving images are taken out and illegally copied, and the illegally copied image is an I picture, that is, an intra-frame coding system. In view of the above, attention is paid to the fact that the digital watermark detection process can be easily performed, and a mechanism for limiting the image that can be output as a still image from the image display apparatus to only the I picture is provided. In other words, even if a user who uses an image display device that displays a moving image performs a pause operation of the moving image, a restriction mechanism is provided so that the moving image cannot be paused except for an I-picture frame. As a result, copyright protection can be surely performed.

  On the other hand, when performing digital watermark detection, the probability of whether or not the watermark data embedded in an image has been detected correctly is generally called detection accuracy. In order to increase this detection accuracy, in general, A method of simply increasing the number of pixels to be changed and the amount of change with respect to the amount of data to be embedded is used. However, this method has a problem that image quality is expected to decrease in proportion to an increase in data to be embedded.

  In the present invention described below, it is assumed that a technique known and described in Patent Document 3 is used as an algorithm for embedding and detecting a digital watermark. The algorithm described in Patent Document 3 virtually divides an image into a plurality of rectangular blocks when embedding a digital watermark, embeds luminance noise at the same relative position in each rectangular block, and detects an electronic watermark, This noise is statistically detected and taken out as watermark information.

  The above algorithm can detect watermark information embedded in an image with flat luminance with high accuracy, but watermark information embedded in an image with random luminance tends to have low accuracy when detected. It has the characteristics.

On the other hand, human vision has a characteristic that noise in a flat portion of an image is easily perceived and insensitive to noise in a messy portion. In addition to this, human vision easily perceives edge noise such as the outline of a face, and also has a characteristic of not recognizing a fine part of a moving part as a moving picture. When embedding digital watermarks, embedding a large amount of watermark information while minimizing the degradation of image quality perceived by humans by embedding in consideration of the characteristics of human vision as described above. Is possible.
US Pat. No. 5,636,292 JP 2002-10224 A JP 2000-270205 A

  The quality of a P picture image in an MPEG video stream that is often used at present is slightly inferior to that of an I picture image, but does not change significantly. On the other hand, the quality of a B picture image is significantly inferior to the quality of other picture images. This is because the I picture and P picture are images that are referred to as prediction frames by the inter-frame coding scheme, whereas the B picture is used only for display and is referred to as a prediction frame. Because it is not done. In the case of an image that is referred to as a prediction frame, the quality of the image directly affects the image quality of the reference destination, and therefore it is necessary to keep the quality as high as possible. As described above, a moving image compressed by MPEG generally has a difference in image quality depending on a frame.

  However, since the conventional technology described above embeds the same watermark data in all frames and attempts to detect the watermark data from all frames, the efficiency of digital watermark information detection Has the problem of being bad. For example, when watermark data is embedded in the same method for an image compressed as an I picture and an image compressed as a B picture, depending on the content of the moving image, the watermark data is correctly output from the I picture. Even if it can be extracted, it may not be correctly extracted from the B picture, or the detection accuracy of the extracted watermark data may be too low. In such a case, the calculation amount for extracting the watermark data from the B picture is wasted as a result, and in some cases, erroneous information is added to the extracted watermark data. If there is a possibility that wrong information will be carried, it will be necessary to consider later whether the information of the extracted watermark data is wrong, which will further increase the amount of calculation. become.

  An object of the present invention is to detect digital watermark information that solves the above-described problems of the prior art and can efficiently detect highly accurate digital watermark information from an image or video stream subjected to data compression such as MPEG. It is to provide a method and apparatus.

  According to the present invention, in the electronic watermark information detection method for detecting digital watermark information by decoding compressed and encoded image data embedded with digital watermark information, the quantization scale is determined from an input stream. A picture having a small value is determined and selected, and electronic watermark information is detected using a block at a specific position in the selected picture. Also, an average value of a quantization scale is calculated for each picture of an input stream. This is achieved by calculating, determining and selecting a picture having a small quantization scale of the calculated average value, and detecting electronic watermark information using the selected picture.

  In addition, the object is achieved by accumulating a plurality of selected pictures, using a block at a specific position in the added picture, and detecting electronic watermark information using the added picture. Is done.

  Another object of the present invention is to provide a digital watermark information detecting device for detecting digital watermark information by decoding compressed and encoded image data in which digital watermark information is embedded. A quantization scale for each picture of the input stream by means for determining and selecting a picture possessed and means for detecting electronic watermark information using a block at a specific position in the selected picture. This is achieved by comprising means for calculating an average value, means for determining and selecting a picture having a quantization scale with a small calculated average value, and means for detecting electronic watermark information using the selected picture. .

  According to the present invention, when detecting information such as a digital watermark embedded in advance from a moving image stream compressed by a data compression method such as MPEG, it is possible to reduce the amount of calculation for detection. In addition, it is possible to improve the detection accuracy of the detected digital watermark data and the like, and it is possible to obtain an effect that the image quality can be improved if the detection accuracy is comparable.

  Embodiments of a digital watermark information detection method and apparatus according to the present invention will be described below in detail with reference to the drawings.

Before describing the embodiment of the present invention, first, the basic concept of the present invention will be described. That is, the present invention focuses on the quantization scale (QS) and limits the portion or frame of the image from which the digital watermark is detected. A parameter called quantization scale (QS) is used when performing a quantization operation for compression by MPEG.
Data after quantization = (discrete cosine transform coefficient × 8) ÷ (quantization matrix × 2 × QS)
However, the compression ratio can be adjusted by adjusting the quantization scale.

  This quantization scale is determined to be able to be specified in units of macroblocks (16 dots × 16 dots). For example, it is possible to reduce the compression rate only for complex pattern portions on the screen. It is. If the QS is reduced, the compression rate is reduced and the deterioration of the image is reduced. This means that the detection rate can be increased by detecting a digital watermark using a small QS. In addition, the compression rate can be adjusted by QS even between screens, and a screen with a small QS value has a low compression rate and little deterioration, so that it is easy to detect a digital watermark. The present invention uses such QS characteristics to limit portions or frames in an image with a small QS and detect a digital watermark from them. In the embodiment of the present invention described below, electronic watermark information is embedded in all input images.

  FIG. 1 is a block diagram showing the configuration of a digital watermark information detection apparatus according to the first embodiment of the present invention. In FIG. 1, 1 is a quantization scale size determination unit, 2 is an MPEG decoding unit, 3 is a switch, 4 is a digital watermark extraction unit, 5 is a digital watermark information detection device, 6 is an MPEG stream input device, and 7 and 8 are It is a display device.

  The digital watermark information detection apparatus 5 according to the first embodiment of the present invention receives the MPEG stream input 101, determines the size of the quantization scale, and indicates whether or not to detect the watermark. A quantization scale magnitude determination unit 1 that outputs a determination signal, an MPEG decoding unit 2 that receives an MPEG stream input 101 and decodes the input MPEG signal, and an electronic watermark output from the quantization scale size determination unit 1. The switch 3 is controlled by a QS magnitude determination signal 102 indicating whether or not to detect, and an electronic watermark extracting unit 4 included in a signal input through the switch 3.

  The digital watermark information detection apparatus 5 outputs an MPEG stream embedded with the digital watermark information recorded on a recording medium or distributed wirelessly or by wire to the digital watermark information detection apparatus 5. An input MPEG stream input device 6 is connected, a display device 7 for monitoring the extracted electronic watermark data, and a display device as an image output device for displaying an image decoded by the MPEG decoding unit 2 8 is connected.

  In the digital watermark information detection apparatus 5 configured as described above according to the first embodiment of the present invention, the MPEG stream input 101 input from the external MPEG stream input apparatus 6 is the quantization scale magnitude determination unit 1 and It is supplied to the MPEG decoding means 2. The quantization scale magnitude determination unit 1 determines the magnitude of the QS value in the input stream. This determination is performed by extracting, as statistical information, whether or not specific noise is present at a specific relative position by statistical processing in a single frame (one image). Then, the quantization scale magnitude determination unit 1 sends to the switch 3 a QS magnitude determination signal 102 that is information as to whether or not to detect a digital watermark, as a QS value magnitude determination result.

  The switch 3 operates based on an instruction by the QS magnitude determination signal 102 from the quantization scale magnitude judging means 1, and selects and outputs the designated image data. This instruction is an instruction for sending image data to the digital watermark extracting means 4 when the QS value is small. For this reason, the switch 3 is turned on by the signal 102 in which the quantization scale magnitude determination unit 1 has a small QS value and should detect the digital watermark, and decodes the input MPEG stream and outputs it as image data. The image data from the decoding means 2 is transferred to the digital watermark extraction unit 4. The digital watermark extraction unit 4 processes the input picture in units of frames, and whether or not a digital watermark is embedded in a block at a specific relative position of the frame, and if so, what is its value? Is detected.

  The watermark information detected by the digital watermark extraction unit 4 is output to the display device 7 and displayed for monitoring or the like, and the image data decoded by the MPEG decoding unit 2 is displayed as an image output device. It is output to the device 8 and displayed. Here, the detected watermark information and the decoded image are displayed on separate display devices, but the present invention allows them to be displayed simultaneously on one display device. Can do. In this case, the watermark information may be displayed by providing a watermark information display area at a position that does not interfere with the displayed image, for example, at the bottom, top, left side, and right side of the screen. Alternatively, the image may be superimposed on the image displayed at these positions.

  According to the first embodiment of the present invention described above, only the frame that is judged to have high detection accuracy of the digital watermark data is input to the digital watermark extraction unit 4. Information with high accuracy can be obtained as the electronic watermark information output from 4.

  FIG. 2 is a block diagram showing a configuration of a digital watermark information detection apparatus according to the second embodiment of the present invention. In FIG. 2, 9 is a quantization scale average value calculation unit, and other symbols are the same as those in FIG.

  The second embodiment of the present invention shown in FIG. 2 is that a quantization scale average value calculation unit 9 that receives the input MPEG stream 101 is provided before the quantization scale magnitude determination unit 1. The other configurations are the same as those of the first embodiment of the present invention described with reference to FIG.

  The quantization scale average value calculation unit 9 calculates an average value of QS of each frame of the input stream. As described above, the QS is set in units of blocks, and since there are a large number of blocks in one frame, the quantization scale average value calculator 9 calculates the average value of QS set in these blocks. Is calculated. The quantization scale size determination means 1 can determine how much the image quality is less deteriorated compared to other frames by determining the magnitude of the average value calculated above. To control the switch 3. The operation of the other functional units in the second embodiment of the present invention is the same as that in the first embodiment of the present invention described with reference to FIG. The watermark extraction unit 4 extracts electronic watermark information from the entire frame or from a block at a specific relative position of the frame. According to the second embodiment of the present invention, the same effects as those of the first embodiment of the present invention described above can be obtained.

  Next, a third embodiment of the present invention will be described. Prior to the description of the third embodiment, the arrangement of pictures in a standard MPEG stream will be described.

  FIG. 3 is a diagram for explaining an example of the arrangement of pictures in a standardly used MPEG stream. In FIG. 3, 201 is an MPEG stream, 202 is a GOP (Group of Pictures) in the MPEG stream, 211 is an I picture, and 212 is a P picture.

  In the example shown in FIG. 3, the MPEG stream used as a standard constitutes a GOP for every 15 pictures, and one I picture and four P pictures are included in the GOP composed of 15 pictures. They appear to appear every three frames. It should be noted that it is possible to arbitrarily set how many pictures a 1 GOP is composed of and how many frames each have an I picture and a P picture.

  The third embodiment of the present invention accumulates and stores pictures having a small average QS value among a plurality of frames based on the arrangement of pictures in the MPEG stream as described above, and inputs the accumulated pictures to the digital watermark extracting means. This makes it possible to detect electronic watermark with high accuracy.

  FIG. 4 is a block diagram showing a configuration of a digital watermark information detecting apparatus according to the third embodiment of the present invention. In FIG. 4, 11 is a cumulative addition processing unit, 12 is a cumulative information storage unit, and the other symbols are the same as those in FIG.

  In the third embodiment of the present invention shown in FIG. 4, the processing operations in the quantization scale magnitude determination unit 1 and the MPEG decoding unit 2 are the same as those in the first embodiment of the present invention described above. In the example shown in FIG. 4 as well, as in the case of the second embodiment of the present invention described above, the quantization scale average value reception that receives the input MPEG stream is preceded by the quantization scale magnitude determination unit 1. Part 9 may be provided. In the third embodiment of the present invention, the cumulative addition processing unit 11 that adds a plurality of frames transferred through the switch 3 and the cumulative information from the cumulative addition processing unit 11 are temporarily stored and electronically transparent. A cumulative information storage unit 12 to be passed to the detection unit 4 is provided.

  In the third embodiment of the present invention having the configuration as described above, each picture obtained by decoding each frame of the MPEG stream 201 is output as image data from the MPEG decoding means 2 according to the time axis indicated by the arrow in FIG. Is done. The quantization scale magnitude determination unit 1 detects a picture having a small QS average value from a plurality of frames of the MPEG stream 201 and closes the switch 3, and a plurality of these pictures are input to the cumulative addition processing unit 11.

  The cumulative addition processing unit 11 performs a statistical addition process on a picture having a preset number of frames, and performs a process for obtaining, as statistical information, whether or not a specific noise is on a specific relative position on the image data. Do. The result of the addition process is temporarily stored in the accumulated information storage unit 12 as accumulated information, and then sequentially sent to the digital watermark extracting unit 4.

  In the first embodiment of the present invention described above, whether or not a specific noise is present at a specific relative position is extracted as statistical information by statistical processing within a single frame. In the third embodiment, similar statistical processing is performed on a plurality of frames, that is, a larger number of sample images, and whether or not noise specific to a specific relative position is present is extracted as statistical information. Thereby, the third embodiment of the present invention can detect a digital watermark with high detection accuracy. In particular, the third embodiment of the present invention detects electronic watermark in an image in which an MPEG stream is a still image, that is, an image that is displayed for a certain period of time, for example, even if the image is a moving image. In some cases, watermark information with extremely high accuracy can be obtained.

  As a method of generating cumulative picture information, information for each picture is stored separately and addition processing is performed as necessary, or addition processing is sequentially performed every time a picture to be accumulated appears. Can take the way. Further, it is not necessary to process the accumulated statistical information and the statistical information from the current detection target picture with the same weighting, for example, processing that places importance on the information from the current detection target picture. May be.

  In the above description, pictures having a small QS average value input to the cumulative addition processing unit 11 under the control of the quantization scale magnitude determination unit 1 are an I picture and a P picture, but a B picture also has a QS average value. There are small ones, and such B pictures can be included. That is, the pictures input to the cumulative addition processing unit 11 can be only I pictures, or can be I pictures and P pictures. Furthermore, a picture with a small average QS value in a B picture can be added to the I picture and the P picture and input to the cumulative addition processing unit 11.

  According to the third embodiment of the present invention described above, pictures having a small average QS value among a plurality of frames based on the arrangement of pictures in the MPEG stream are accumulated and stored and input to the digital watermark extracting means. Therefore, it is possible to detect the electronic watermark with higher accuracy.

It is a block diagram which shows the structure of the detection apparatus of the digital watermark information by the 1st Embodiment of this invention. It is a block diagram which shows the structure of the detection apparatus of the digital watermark information by the 2nd Embodiment of this invention. It is a figure explaining the example of the arrangement | sequence of the picture in the MPEG stream used normally. It is a block diagram which shows the structure of the detection apparatus of the digital watermark information by the 3rd Embodiment of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Quantization scale magnitude determination part 2 MPEG decoding part 3 Switch 4 Digital watermark extraction part 5 Digital watermark information detection apparatus 6 MPEG stream input apparatus 7, 8 Display apparatus 9 Quantization scale average value calculation part 11 Cumulative addition process part 12 Cumulative Information storage unit 201 MPEG stream 202 GOP (Group of Pictures) in MPEG stream
211 I picture 212 P picture

Claims (14)

  A digital watermark information detection method that detects digital watermark information by decoding compressed and encoded image data embedded with digital watermark information, and determines and selects a picture with a small quantization scale from the input stream An electronic watermark information detecting method, wherein electronic watermark information is detected using a block at a specific position in a selected picture.   In a digital watermark information detection method for detecting digital watermark information by decoding digitally encoded image data embedded with digital watermark information, calculating an average value of a quantization scale for each picture of an input stream, A digital watermark information detection method comprising: judging and selecting a picture having a quantization scale with a small calculated average value, and detecting digital watermark information using the selected picture.   A digital watermark information detection method that detects digital watermark information by decoding compressed and encoded image data embedded with digital watermark information, and determines and selects a picture with a small quantization scale from the input stream A digital watermark information detection method comprising: adding a plurality of selected pictures cumulatively, and detecting digital watermark information using a block at a specific position in the added picture.   In a digital watermark information detection method for detecting digital watermark information by decoding digitally encoded image data embedded with digital watermark information, calculating an average value of a quantization scale for each picture of an input stream, Digital watermark information detection characterized by judging and selecting a picture having a small quantization scale with a calculated average value, accumulating a plurality of selected pictures, and detecting digital watermark information using the added pictures Method.   5. The digital watermark information detection method according to claim 1, wherein the compression-encoded image data is image data compression-encoded by MPEG.   4. The compression-coded image data is image data compressed and encoded by MPEG, and the pictures to be cumulatively added are I pictures, or I pictures and P pictures. Or the digital watermark information detecting method according to 4.   The compression-encoded image data is image data compression-encoded by MPEG, and the pictures to be cumulatively added are an I picture, a P picture, and a B picture having a small quantization scale. The digital watermark information detecting method according to claim 3 or 4.   A digital watermark information detection device that detects digital watermark information by decoding compressed and encoded image data with embedded digital watermark information, and determines and selects a picture with a small quantization scale from the input stream An electronic watermark information detecting apparatus comprising: means for performing electronic watermark information detection using a block at a specific position in a selected picture.   Means for calculating an average value of a quantization scale for each picture of an input stream in a digital watermark information detecting apparatus for detecting digital watermark information by decoding compression-encoded image data embedded with digital watermark information And a means for determining and selecting a picture having a calculated quantization scale with a small average value, and means for detecting digital watermark information using the selected picture.   A digital watermark information detection device that detects digital watermark information by decoding compressed and encoded image data with embedded digital watermark information, and determines and selects a picture with a small quantization scale from the input stream A digital watermark information detecting apparatus comprising: means for performing cumulative addition of a plurality of selected pictures; and means for detecting digital watermark information using a block at a specific position in the added picture.   Means for calculating an average value of a quantization scale for each picture of an input stream in a digital watermark information detecting apparatus for detecting digital watermark information by decoding compression-encoded image data embedded with digital watermark information And means for determining and selecting a picture having a small quantization scale with a calculated average value, means for cumulatively adding a plurality of selected pictures, and means for detecting electronic watermark information using the added pictures An electronic watermark information detection apparatus comprising:   12. The digital watermark information detection apparatus according to claim 8, wherein the compression-coded image data is image data compression-coded by MPEG.   11. The compression-coded image data is image data compression-coded by MPEG, and the pictures to be cumulatively added are I pictures or I pictures and P pictures. Or the electronic watermark information detecting apparatus according to 11.   The compression-encoded image data is image data compression-encoded by MPEG, and the pictures to be cumulatively added are an I picture, a P picture, and a B picture having a small quantization scale. The digital watermark information detecting method according to claim 10 or 11.

Images