Geometric attack resistant watermarking algorithm

Abstract

The invention provides a geometric attack resisting watermark algorithm. The geometric attack resisting watermark algorithm comprises the following steps: dividing an original image into Z sub-image blocks; performing error correction coding on the original watermark information to obtain coded watermark information; dispersing and embedding the B-bit coded watermark information into the Z sub-image blocks to obtain a watermarked image; after the watermarked image is subjected to geometric attack, obtaining a watermarked image after geometric transformation; the geometric attack comprises cutting and / or scale amplification; resampling the image with the watermarks after geometric transformation to obtain a restored image with the watermarks; extracting the coded watermark information to obtain a coded watermark; and performing error correction decoding on the coded watermark to obtain original watermark information. According to the method, coded watermark information is dispersed and embedded into sub-image blocks of an original image, resampling is carried out in a boundary filling expansion or pixel value mapping reduction mode, a restored watermarked image is obtained, and original watermark information is obtained through error correction decoding. The geometric attack resisting watermark algorithm is high in precision, low in cost and high in generalization.

Description

Technical Field

[0001] The invention belongs to the technical field of image processing, and in particular relates to a watermark algorithm resistant to geometric attacks. Background Art

[0002] Digital watermarking technology is widely used in copyright protection, content verification and other fields. Traditional watermarking methods embed hidden information in the frequency domain of digital media to achieve the purpose of content protection. Geometric attack is a common image transformation method, which is widely used in image transmission processes, such as cropping and scaling.

[0003] Traditional watermarking methods have good resistance to point-to-point image transformations such as compression, and can maintain a high watermark extraction accuracy. However, for geometric attacks, the watermark extraction accuracy will be seriously reduced. Existing anti-geometric attack methods are complex in calculation, low in precision, and have a single type of anti-attack. Summary of the invention

[0004] The purpose of the present invention is to provide a geometric attack resistant watermarking algorithm, which embeds the encoded watermark information in the sub-image blocks of the original image in a dispersed manner, and resamples by enlarging the boundary by filling or reducing the pixel value by mapping, so as to obtain the restored watermarked image. The geometric attack resistant watermarking algorithm of the present invention has high precision, low cost and strong generalization.

[0005] The present invention provides a watermark algorithm resistant to geometric attacks, comprising:

[0006] S1, dividing the original image into Z sub-image blocks;

[0007] S2, performing error correction coding on the original watermark information to obtain coded watermark information; the total number of binary digits of the coded watermark information is B, where B is the sum of the number of watermark information bits and the number of coding check bits;

[0008] S3, dispersively embedding the encoded watermark information of B bits into the Z sub-image blocks to obtain a watermarked image;

[0009] S4, the watermarked image is subjected to a geometric attack to obtain a watermarked image after geometric transformation; the geometric attack includes cropping and / or scaling;

[0010] S5, resampling the geometrically transformed watermarked image by filling and enlarging or pixel value mapping and reducing, so that the geometrically transformed watermarked image is restored to the size of the original image, to obtain a restored watermarked image;

[0011] S6, extracting the encoded watermark information from the sub-image block included in the restored watermarked image to obtain the encoded watermark;

[0012] S7. Perform error correction decoding on the encoded watermark to obtain the original watermark information.

[0013] Furthermore, when the geometric attack is cropping, step S5 specifically includes: filling the geometrically transformed watermarked image obtained by cropping to the size of the original image; and filling the filled area with the same pixel value.

[0014] Furthermore, when the geometric attack is scale magnification, the geometrically transformed watermarked image obtained by scale magnification is defined as an enlarged watermarked image; step S5 specifically includes:

[0015] Obtaining the resolution and scale magnification of the original image;

[0016] Calculate the mapping relationship between the original image and the enlarged watermarked image; find the mapping point coordinates in the enlarged watermarked image corresponding to any original point coordinates in the original image according to the mapping relationship, use the corresponding pixel of the mapping point coordinates as the pixel value of the original point coordinates, traverse all the original point coordinates within the resolution range of the original image, and obtain the pixel values ​​corresponding to all the original point coordinates, so as to obtain the restored watermarked image with the same resolution as the original image.

[0017] Furthermore, the mapping relationship adopts the formula f=[x*scale], where f is the coordinate of the mapping point, x is the coordinate of the original point, scale is the scale magnification, and [] is a rounding operation.

[0018] Furthermore, when the mapping point coordinates are integers, the pixel at the location of the mapping point coordinates is used as the corresponding pixel of the mapping point coordinates;

[0019] When the mapping point coordinates are non-integer, the pixels at the positions where the coordinates of the integer points adjacent to the mapping point coordinates are located are used as the corresponding pixels of the mapping point coordinates.

[0020] Furthermore, the geometric attack includes: any one of single cropping, single scale transformation, cropping first and then scale transformation, and scale transformation first and then cropping.

[0021] Furthermore, step S3 specifically includes: calculating the number K of the sub-image blocks into which each bit of the encoded watermark information needs to be embedded, K=Z / B; and embedding each bit of the encoded watermark information into K sub-image blocks that are roughly evenly distributed in the original image.

[0022] Furthermore, in step S3, the sequence numbers of the 1st to Zth sub-image blocks are randomly arranged to obtain a random sequence; and the K sub-image blocks corresponding to each bit of the encoded watermark information are successively configured into the random sequence;

[0023] Specifically, the first group of K sub-image blocks from front to back in the random sequence are respectively embedded with the first bit of the encoded watermark information; the second group of K sub-image blocks in the random sequence are respectively embedded with the second bit of the encoded watermark information; and so on, the Bth group of K sub-image blocks in the random sequence are respectively embedded with the Bth bit of the encoded watermark information.

[0024] Furthermore, the error correction coding includes: BCH coding or Hamming code.

[0025] Furthermore, the original watermark information length is Y bits, wherein each m-bit watermark information bit is a watermark information group, and each of the watermark information groups is correspondingly configured with a coding check group containing n-bit coding check bits; B = [Y / m]*(m+n), [] is a round-up operator.

[0026] Compared with the prior art, the present invention has the following beneficial effects:

[0027] The present invention provides a watermarking algorithm resistant to geometric attacks, comprising: dividing an original image into Z sub-image blocks; performing error correction coding on the original watermark information to obtain coded watermark information; the total number of binary bits of the coded watermark information is B, where B is the sum of the number of bits of the watermark information bits and the number of bits of the coding check bits; distributing and embedding the B-bit coded watermark information into the Z sub-image blocks to obtain a watermarked image; after the watermarked image is subjected to a geometric attack, a geometrically transformed watermarked image is obtained; the geometric attack includes cropping and / or scale enlargement; resampling the geometrically transformed watermarked image by filling and enlarging or pixel value mapping and reducing, so that the geometrically transformed watermarked image is restored to the size of the original image to obtain a restored watermarked image; extracting the coded watermark information from the sub-image blocks contained in the restored watermarked image to obtain the coded watermark; performing error correction decoding on the coded watermark to obtain the original watermark information. The present invention distributively embeds the coded watermark information into the sub-image blocks of the original image, and resamples by boundary filling and enlarging or pixel value mapping and reducing to obtain a restored watermarked image. The geometric attack resistant watermark algorithm of the present invention has high precision, low cost and strong generalization. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 The figure is a schematic diagram of the flow of the anti-geometric attack watermark algorithm according to an embodiment of the present invention.

[0029] Figure 2 It is a schematic diagram of the encoded watermark information obtained after error correction encoding in an embodiment of the present invention.

[0030] Figure 3 The figure is a schematic diagram of mapping between each bit of encoded watermark information and the corresponding embedded random sub-image block sequence number in an embodiment of the present invention.

[0031] Figure 4 The first bit of the encoded watermark information B in the embodiment of the present invention 0 Schematic diagram of distributed embedding into the first set of K random sub-image blocks.

[0032] Figure 5 This is a schematic diagram of obtaining a restored watermarked image by resampling the geometrically transformed watermarked image through filling and enlarging after a cropping attack in an embodiment of the present invention.

[0033] Figure 6 Schematic diagram of comparison of watermark bit error rate (BER) measured under three conditions: no cropping, first cropping size, and second cropping size.

[0034] Figure 7 The figure is a comparison diagram of the watermark bit error rate (BER) measured under three conditions: no amplification, the first integer amplification factor, and the second integer amplification factor.

[0035] Figure 8 The figure is a comparison diagram of the watermark bit error rate (BER) measured without amplification and under three different floating-point amplification factors.

[0036] Fig. 9 Schematic diagram of the comparison of watermark bit error rate (BER) measured without geometric attack and under different cropping and scale magnification combinations. DETAILED DESCRIPTION

[0037] The present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. The advantages and features of the present invention will become more apparent from the following description. It should be noted that the accompanying drawings are in a very simplified form and use an inaccurate scale, which is only used to conveniently and clearly assist in explaining the purpose of the embodiments of the present invention.

[0038] For ease of description, some embodiments of the present application may use spatially relative terms such as "above", "below", "top", "below", etc. to describe the relationship between one element or component and another (or other) elements or components as shown in the various figures of the embodiments. It should be understood that in addition to the orientations described in the drawings, the spatially relative terms are also intended to include different orientations of the device in use or operation. For example, if the device in the drawings is turned over, the elements or components described as being "below" or "below" other elements or components will subsequently be positioned as being "above" or "above" other elements or components. The terms "first", "second", etc. below are used to distinguish between similar elements and are not necessarily used to describe a specific order or time sequence. It is to be understood that these terms used in this way are interchangeable where appropriate.

[0039] The embodiment of the present invention provides a watermark algorithm resistant to geometric attacks, such as Figure 1As shown, including:

[0040] Step S1, dividing the original image into Z sub-image blocks;

[0041] Step S2, performing error correction coding on the original watermark information to obtain coded watermark information; the total number of binary digits of the coded watermark information is B, where B is the sum of the number of watermark information bits and the number of coding check bits;

[0042] Step S3, embedding the B-bit encoded watermark information into Z sub-image blocks in a distributed manner to obtain a watermarked image;

[0043] Step S4, the watermarked image is subjected to a geometric attack to obtain a watermarked image after geometric transformation; the geometric attack includes cropping and / or scaling;

[0044] Step S5, resampling the geometrically transformed watermarked image by filling and enlarging or pixel value mapping and reducing, so that the geometrically transformed watermarked image is restored to the size of the original image, thereby obtaining a restored watermarked image;

[0045] Step S6, extracting the encoded watermark information from the sub-image blocks included in the restored watermarked image to obtain the encoded watermark;

[0046] Step S7: perform error correction decoding on the encoded watermark to obtain the original watermark information.

[0047] The steps of the anti-geometric attack watermark algorithm of the present invention are described in detail below with reference to the accompanying drawings.

[0048] Step S1: Divide the original image into Z sub-image blocks.

[0049] Figure 2 FIG. 1 is a schematic diagram of the encoded watermark information obtained after error correction encoding in an embodiment of the present invention. Figure 2 As shown, the original watermark information is error-corrected and encoded to obtain the encoded watermark information; the total number of binary bits of the encoded watermark information is B, where B is the sum of the number of watermark information bits and the number of coding check bits. Error correction coding includes, for example, BCH coding or Hamming code. The length of the original watermark information is Y bits, where each m-bit watermark information bit is a watermark information group, and each watermark information group is configured with a coding check group containing n-bit coding check bits; B = [Y / m]*(m+n), [] is a round-up operator.

[0050] Figure 3 The figure is a schematic diagram of mapping between each bit of encoded watermark information and the corresponding embedded random sub-image block sequence number in an embodiment of the present invention. Figure 4 The first bit of the encoded watermark information B in the embodiment of the present invention 0Schematic diagram of distributing and embedding into the first group of K random sub-image blocks. Step S3: Figure 3 and Figure 4 As shown, B bits of encoded watermark information are dispersedly embedded into Z sub-image blocks to obtain a watermarked image. Exemplarily, each bit of encoded watermark information is embedded into at least one sub-image block.

[0051] Specifically, calculate the number K of sub-image blocks that need to be embedded into each bit of the encoded watermark information, K = Z / B, where the total number of sub-image blocks is Z and the total number of bits of the encoded watermark information is B. Each bit of the encoded watermark information is repeatedly embedded into K sub-image blocks that are roughly evenly distributed in the original image. Exemplarily, the sequence numbers of the 1st to Zth sub-image blocks are randomly arranged to obtain a random sequence. Each bit of the encoded watermark information corresponds to K sub-image blocks, which are successively configured in the random sequence. The first group of K sub-image blocks from the front to the back of the specific random sequence are respectively embedded with the first bit of the encoded watermark information B 0 The second group of K sub-image blocks in the random sequence are respectively embedded with the second bit of encoded watermark information B 1 ; Similarly, the B-th group of K sub-image blocks in the random sequence are respectively embedded with the B-th bit of encoded watermark information.

[0052] Step S4: After the watermarked image is subjected to a geometric attack, a watermarked image after geometric transformation is obtained; the geometric attack includes cropping and / or scaling.

[0053] Step S5, resampling the geometrically transformed watermarked image by filling and enlarging or pixel value mapping and reducing, so that the geometrically transformed watermarked image is restored to the size of the original image, thereby obtaining a restored watermarked image.

[0054] Figure 5 This is a schematic diagram of the restored watermarked image obtained by resampling the geometrically transformed watermarked image by filling and expanding after the cropping attack in an embodiment of the present invention. Figure 5 As shown, when the geometric attack is cropping, step S5 specifically includes: filling the geometrically transformed watermarked image obtained by cropping to the size of the original image. The filled area is filled with the same pixel value. The filled pixel value can be any pixel value from 0 to the maximum pixel value. In one example, the pixel value filled in the filled area is, for example, 0. Cropping can be, for example, equidistant cropping up and down, left and right, etc., or unequal cropping up and down, left and right, etc. The cropping rate is the ratio of the area of ​​the image cropped to the original image.

[0055] Figure 6 The following is a comparison diagram of the watermark bit error rate (BER) measured under three conditions: no cropping, the first cropping size, and the second cropping size. Figure 6As shown in the figure, the cropping attack is tested. The test set includes 70 1280*720 raw images, and the length of the watermark information after encoding is 32Byte. Figure 6 It can be seen that under different cropping methods, the accuracy of this method is almost exactly the same as before cropping. The repeated embedding of the coded watermark information (redundant information) greatly reduces the impact of missing watermarks in the edge area and effectively improves the accuracy. Each bit of coded watermark information is repeatedly embedded in K sub-image blocks that are roughly evenly distributed in the original image. For example, the coded watermark information in the edge area of ​​the watermarked image is cropped, but the remaining area after the watermarked image is cropped still has repeated coded watermark information distribution, so even if it is cropped, the watermark information can still be effectively extracted.

[0056] When the geometric attack is scale magnification, the watermarked image after geometric transformation obtained by scale magnification is defined as the magnified watermarked image; step S5 specifically includes: obtaining the resolution and scale magnification of the original image. Calculate the mapping relationship between the original image and the magnified watermarked image; find the mapping point coordinates in the magnified watermarked image corresponding to any original point coordinate in the original image according to the mapping relationship, take the corresponding pixel of the mapping point coordinates as the pixel value of the original point coordinates, traverse all the original point coordinates within the resolution range of the original image, and obtain the pixel values ​​corresponding to all the original point coordinates, so as to obtain the restored watermarked image with the same resolution size as the original image. The mapping relationship adopts the formula f=[x*scale], where f is the mapping point coordinate, x is the original point coordinate, scale is the scale magnification, and [] is the rounding operation. Scale magnification can be, for example, proportional magnification of the length and width of the image. When the coordinates of the mapping point are integers, the pixels at the location of the mapping point coordinates are used as the corresponding pixels of the mapping point coordinates; when the coordinates of the mapping point are non-integers, the pixels at the location of the coordinates of the neighboring integer points of the mapping point coordinates are used as the corresponding pixels of the mapping point coordinates, that is, the resampling of the present invention adopts fixed-point nearest neighbor mapping, which is low-cost and high-precision. The anti-geometric attack strategy only requires simple fixed-point operations, low computational overhead, and low cost.

[0057] Figure 7 Schematic diagram showing comparison of watermark bit error rates (BER) measured under three conditions: no scale, first integer scale (eg, 2 times), and second integer scale (eg, 4 times). Figure 8 The following is a comparison diagram of the watermark bit error rate (BER) measured without amplification (No scale) and under three different floating-point amplification conditions. Figure 7 and Figure 8 As shown in the figure, the scale-enlargement attack is tested. The test set includes 70 1280*720 raw images, and the length of the watermark information after encoding is 32Byte. Figure 7 and Figure 8It can be seen that under integer and non-integer (floating point) magnification modes, the accuracy of the watermark algorithm resistant to scale magnification geometric attack is very close to that of the non-scale magnification geometric attack. Resampling can effectively restore the original watermark information and resist the influence of scale magnification.

[0058] Fig. 9 The figure is a comparison diagram of the watermark bit error rate (BER) measured under the conditions of no geometric attack, different cropping and scale magnification combinations. Geometric attacks include: any one of single cropping, single scale transformation, cropping first and then scale transformation, and scale transformation first and then cropping. The anti-scale magnification scheme of the resampling of the present invention can be combined with the anti-cropping scheme to resist combined attacks and meet the needs of real applications. The combination of cropping and scale magnification attacks can obtain an accuracy rate that is very close to that when there is no geometric attack (for example, 0.0898) (both are approximately equal to 0.09), achieving high precision. The bit error rate (BER) of the watermark of the present invention is low, and the lower the bit error rate, the higher the accuracy of the extracted watermark information. For different cropping and scale magnification methods (including their combinations), the accuracy rate is consistent and has strong generalization.

[0059] In summary, the present invention provides a geometric attack resistant watermark algorithm, comprising: dividing an original image into Z sub-image blocks; performing error correction coding on the original watermark information to obtain coded watermark information; the total number of binary bits of the coded watermark information is B, where B is the sum of the number of bits of the watermark information bits and the number of bits of the coding check bits; dispersively embedding the B-bit coded watermark information into the Z sub-image blocks to obtain a watermarked image; after the watermarked image is subjected to a geometric attack, a geometrically transformed watermarked image is obtained; the geometric attack includes cropping and / or scale enlargement; resampling the geometrically transformed watermarked image by filling and enlarging or pixel value mapping and reducing, so that the geometrically transformed watermarked image is restored to the size of the original image to obtain a restored watermarked image; extracting the coded watermark information from the sub-image blocks contained in the restored watermarked image to obtain the coded watermark; performing error correction decoding on the coded watermark to obtain the original watermark information. The invention embeds the encoded watermark information in the sub-image blocks of the original image in a dispersed manner, and resamples by enlarging the boundary by filling or reducing the pixel value by mapping, so as to obtain the restored watermarked image. The watermark algorithm resistant to geometric attacks of the invention has high precision, low cost and strong generalization.

[0060] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the method disclosed in the embodiment, since it corresponds to the device disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.

[0061] The above description is only a description of the preferred embodiment of the present invention, and is not any limitation on the scope of rights of the present invention. Any technical personnel in this field can make possible changes and modifications to the technical solution of the present invention by using the methods and technical contents disclosed above without departing from the spirit and scope of the present invention. Therefore, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention shall fall within the protection scope of the technical solution of the present invention.

Claims

1. A watermark algorithm resistant to geometric attacks, characterized in that: include: S1, dividing the original image into Z sub-image blocks; S2, performing error correction encoding on the original watermark information to obtain encoded watermark information; The total number of binary digits of the encoded watermark information is B, where B is the sum of the number of watermark information bits and the number of encoding check bits; S3, dispersively embedding the encoded watermark information of B bits into the Z sub-image blocks to obtain a watermarked image; S4, the watermarked image is subjected to a geometric attack to obtain a watermarked image after geometric transformation; the geometric attack includes cropping and / or scaling; S5, resampling the geometrically transformed watermarked image by filling and enlarging or pixel value mapping and reducing, so that the geometrically transformed watermarked image is restored to the size of the original image, to obtain a restored watermarked image; S6, extracting the encoded watermark information from the sub-image block included in the restored watermarked image to obtain the encoded watermark; S7. Perform error correction decoding on the encoded watermark to obtain the original watermark information.

2. The geometric attack resistant watermark algorithm as claimed in claim 1, characterized in that: When the geometric attack is cropping, step S5 specifically includes: filling the geometrically transformed watermarked image obtained by cropping to the size of the original image; and filling the filled area with the same pixel value.

3. The geometric attack resistant watermark algorithm as claimed in claim 1, characterized in that: When the geometric attack is scale magnification, the geometrically transformed watermarked image obtained by scale magnification is defined as an enlarged watermarked image; Step S5 specifically includes: Obtaining the resolution and scale magnification of the original image; Calculate the mapping relationship between the original image and the enlarged watermarked image; find the mapping point coordinates in the enlarged watermarked image corresponding to any original point coordinates in the original image according to the mapping relationship, use the corresponding pixel of the mapping point coordinates as the pixel value of the original point coordinates, traverse all the original point coordinates within the resolution range of the original image, and obtain the pixel values ​​corresponding to all the original point coordinates, so as to obtain the restored watermarked image with the same resolution as the original image.

4. The geometric attack resistant watermark algorithm as claimed in claim 3, characterized in that: The mapping relationship adopts the formula f=[x*scale], where f is the coordinate of the mapping point, x is the coordinate of the original point, scale is the scale magnification, and [] is a rounding operation.

5. The geometric attack resistant watermark algorithm as claimed in claim 3, characterized in that: When the mapping point coordinates are integers, the pixel at the location of the mapping point coordinates is used as the corresponding pixel of the mapping point coordinates; When the mapping point coordinates are non-integer, the pixels at the positions where the coordinates of the integer points adjacent to the mapping point coordinates are located are used as the corresponding pixels of the mapping point coordinates.

6. The geometric attack resistant watermark algorithm as claimed in claim 1, characterized in that: The geometric attack includes: any one of single cropping, single scale transformation, cropping first and then scale transformation, and scale transformation first and then cropping.

7. The geometric attack resistant watermark algorithm as claimed in claim 1, characterized in that: Step S3 specifically includes: calculating the number K of the sub-image blocks into which each bit of the encoded watermark information needs to be embedded, K=Z / B; and embedding each bit of the encoded watermark information into K sub-image blocks that are roughly evenly distributed in the original image.

8. The geometric attack resistant watermark algorithm as claimed in claim 7, characterized in that: In step S3, the serial numbers of the 1st to Zth sub-image blocks are randomly arranged to obtain a random sequence; and K sub-image blocks are successively arranged into the random sequence according to each bit of the encoded watermark information corresponding to each K sub-image block; Specifically, the first group of K sub-image blocks from the front to the back of the random sequence are respectively embedded with the first bit of the encoded watermark information; The second group of K sub-image blocks in the random sequence are respectively embedded with the second bit of the encoded watermark information; and by analogy, the Bth group of K sub-image blocks in the random sequence are respectively embedded with the Bth bit of the encoded watermark information.

9. The geometric attack resistant watermark algorithm as claimed in claim 1, characterized in that: The error correction code includes: BCH code or Hamming code.

10. The geometric attack resistant watermark algorithm according to claim 1, characterized in that: The length of the original watermark information is Y bits, wherein each m-bit watermark information bit is a watermark information group, and each of the watermark information groups is correspondingly configured with a coding check group containing n-bit coding check bits; B=[Y / m]*(m+n), [] is a round-up operator.

Images