Active mode digital image content identification method based on wavelet and DCT dual domain

Abstract

This invention discloses one active digital image content identification method based on wavelet and DCT double areas, which comprises the following steps: send end processes water mark generation, imbedding and image identification and modification initial position, wherein, through above position to generate water mark images to restore property extraction and imbedding water mark and restoring.

Description

technical field

[0001] The invention relates to an active digital image content identification method, which is an active digital image content identification method based on wavelet and DCT dual domains. Background technique

[0002] The convenience and insecurity of digital images go hand in hand. The digitalization of images provides great efficiency for its transmission and editing, but it also brings a series of security risks: the illegal tampering and dissemination of digital images becomes easy without leaving traces, and these data involve financial , e-commerce and military information communication and many other aspects, so the implementation of effective content protection for digital images in the network environment has become an urgent and realistic problem. In order to solve this problem, content-based digital image identification technology emerges as the times require, and it is not only a legal issue, but also a very important technical issue.

[0003] ...

Examples

Embodiment 1

[0123] As shown in Figure 5: the wavelet transform used in the experiment is biorthogonal 9 / 7 wavelet. When embedding the positioning watermark, the size of n=2×2 is selected for the number of wavelet coefficients to be averaged. Both the positioning block and the restoration result are accurate to the order of 8×8. In binary coding, valid=11, that is, the first 11 coefficients after the Z scan are reserved for the DCT quantized coefficients, so that each 8×8 data block is just represented by 64 bits, and the L matrix used to encode the quantized coefficients is :

[0124] L = 7 7 7 5 4 3 2 1 7 6 ...

Embodiment 2

[0134] The malicious tampering mentioned in the present invention mainly refers to that the image has been cut and replaced. As shown in Figure 8, where (a) is the original lena image of 256*256*8; (b) is the watermark image, PSNR=36.18db; (c) is the watermark image after lena’s face has been maliciously tampered with, the following will Use this falsified image to test the image identification method of the present invention.

[0135] The 8(c) image is identified, and the identification result is shown in Figure 9, where (a) is the result of positioning using the positioning watermark embedded in HL2; (b) is positioning using the positioning watermark embedded in LH2 Results; (c) is the initial positioning result of the tampered image formed by merging the positioning results of (a) and (b). It can be seen from (c) that the positioning result is still not good at this time, so the diagonal The expansion operation is very necessary; (d) is the result of one diagonal expansion...

Embodiment 3

[0138] In the following, malicious tampering and accidental distortion are applied to the watermark image at the same time, the purpose is to test whether the authentication method of the present invention can tolerate accidental distortion and detect malicious tampering at the same time. Figure 11(a) is the original sailboat image; (b) is the sailboat watermark image, with a PSNR of 36.02db; (c) is the copy of the boat in the sailboat watermark image, and the scene around the tampered boat is replaced by the surrounding water , and then JPEG compression with an image quality of 60 was performed, and the identification result was that the image suffered malicious tampering and accidental distortion at the same time, and then the positioning operation was performed; (d) is the result of positioning through the positioning watermark embedded in HL2; (e) is the result of positioning through The positioning result of the positioning watermark embedded in LH2; (f) is the initial pos...

technical field

[0001] The invention relates to an active digital image content identification method, which is an active digital image content identification method based on wavelet and DCT dual domains. Background technique

[0002] The convenience and insecurity of digital images go hand in hand. The digitalization of images provides great efficiency for its transmission and editing, but it also brings a series of security risks: the illegal tampering and dissemination of digital images becomes easy without leaving traces, and these data involve financial , e-commerce and military information communication and many other aspects, so the implementation of effective content protection for digital images in the network environment has become an urgent and realistic problem. In order to solve this problem, content-based digital image identification technology emerges as the times require, and it is not only a legal issue, but also a very important technical issue.

[0003] ...

Examples

Embodiment 1

[0123] As shown in Figure 5: the wavelet transform used in the experiment is biorthogonal 9 / 7 wavelet. When embedding the positioning watermark, the size of n=2×2 is selected for the number of wavelet coefficients to be averaged. Both the positioning block and the restoration result are accurate to the order of 8×8. In binary coding, valid=11, that is, the first 11 coefficients after the Z scan are reserved for the DCT quantized coefficients, so that each 8×8 data block is just represented by 64 bits, and the L matrix used to encode the quantized coefficients is :

[0124] L = 7 7 7 5 4 3 2 1 7 6 ...

Embodiment 2

[0134] The malicious tampering mentioned in the present invention mainly refers to that the image has been cut and replaced. As shown in Figure 8, where (a) is the original lena image of 256*256*8; (b) is the watermark image, PSNR=36.18db; (c) is the watermark image after lena’s face has been maliciously tampered with, the following will Use this falsified image to test the image identification method of the present invention.

[0135] The 8(c) image is identified, and the identification result is shown in Figure 9, where (a) is the result of positioning using the positioning watermark embedded in HL2; (b) is positioning using the positioning watermark embedded in LH2 Results; (c) is the initial positioning result of the tampered image formed by merging the positioning results of (a) and (b). It can be seen from (c) that the positioning result is still not good at this time, so the diagonal The expansion operation is very necessary; (d) is the result of one diagonal expansion...

Embodiment 3

[0138] In the following, malicious tampering and accidental distortion are applied to the watermark image at the same time, the purpose is to test whether the authentication method of the present invention can tolerate accidental distortion and detect malicious tampering at the same time. Figure 11(a) is the original sailboat image; (b) is the sailboat watermark image, with a PSNR of 36.02db; (c) is the copy of the boat in the sailboat watermark image, and the scene around the tampered boat is replaced by the surrounding water , and then JPEG compression with an image quality of 60 was performed, and the identification result was that the image suffered malicious tampering and accidental distortion at the same time, and then the positioning operation was performed; (d) is the result of positioning through the positioning watermark embedded in HL2; (e) is the result of positioning through The positioning result of the positioning watermark embedded in LH2; (f) is the initial pos...

Images