CN102142129A - Visual secret sharing-based image content tampering detection method - Google Patents

Abstract

The invention discloses a visual secret sharing-based image content tampering detection method. The method comprises the following steps of: generating a watermarking image by using copyright information of the image; generating a verification shadow and a detection shadow by combining the content characteristic of the image; recovering a water mark from the verification shadow and the detection shadow by using a visual secret sharing technology; and comparing the recovered water mark with the original watermarking to determine whether the image is tampered so as to realize tampering detection and determine the position of a tampered region. In the method, the watermarking image is generated by using the copyright information of the image without embedding and extracting the watermarking, but the tampering detection of image content is realized by using the visual secret sharing technology on the premise of not damaging the original image data.

Description

A kind of based on the secret picture material altering detecting method of sharing of vision

Technical field

The invention belongs to the multi-media information security technical field, be specifically related to a kind of based on the secret picture material altering detecting method of sharing of vision.

Background technology

Along with developing rapidly of network technology and digitizing technique, the interchange of multimedia messages has reached unprecedented range.The application of the development of while image processing techniques and powerful image editing software makes domestic consumer can edit, distort raw image data easily, makes the effect of mixing the spurious with the genuine.This has reduced the credibility of multi-medium data itself to a great extent, has overturned people's's in the eyes of " seeing is believing " traditional concept for a long time.How to realize the popular research topic in content integrity authentication the becoming multi-media information security field of digital picture.

Image digital watermark is a kind of image integrity commonly used authentication and copyright protection technology.As number of patent application is 200610036388.6 to have announced " a kind of reversible water mark method that is used for image authentication ", high-frequency sub-band histogram by revising image is with the SHA-256 cryptographic hash of original image and be used for the image data recovered and embed high-frequency sub-band, by the lowest bit plane of replacing low frequency sub-band significant watermark is embedded low frequency sub-band, distort detection and location according to the watermark of watermark of extracting or extraction and the differential chart of former watermark, relatively the cryptographic hash of cryptographic hash of Ti Quing and image restored is carried out the content integrity checking.Yet, use traditional digital watermark, need algorithm for design that watermark information is embedded in the original artwork, if containing much information of embedding influences visual effect; If the quantity of information that embeds is little, then relatively be easy to extract, do not reach the security of expection.So, not only when embedding and extract, increased computation complexity, and, in image, embedded watermark, changed the raw information of image, destroyed the integrality of original artwork to a certain extent.Therefore, how neither embedding any information in image, can realize content integrity authentication and copyright protection to original artwork again, is a challenging research topic.

In order to address this problem, people such as Chuen-Ching Wang have proposed a New Scheme of Image Watermarking of learning (Visual Cryptography) based on visual cryptography, its basic thought is different from traditional watermarking project, no longer need watermark is embedded in the original image, but utilize secret (the VSS:Visual secret sharing) technology of sharing of vision that watermarking images is divided into two parts, the information of wherein a part of watermark is combined with the main information of original image, obtain an image that contains watermark; The recovery stage is to recover that a part of watermark information from the image that contains watermark, overlaps on another part watermark information, thus the entitlement of authentication image.Philtrums such as Hsieh have proposed a gray level image copyright protection scheme of sharing based on secret; this scheme is utilized wavelet transform to extract the feature of original image and it is combined with watermark information; generate the shadow image, do not need to design the embedding algorithm of watermark equally.2005, they expanded to the coloured image copyright protection to this method again.Above scheme need not embed any information in original artwork, can realize image copyright protection, compares with traditional water mark method, has reduced computation complexity and has kept the integrality of original image.But above method all can only realize copyright protection, does not have the integrated authentication function of image.In addition, also there is shortcoming in various degree respectively, people's such as Chuen-Ching Wang operation is carried out on the spatial domain, its shortcoming is a little less than the image processing operations robustness to routine, people's such as Hsieh method all is on transform domain image to be handled, but the algorithm more complicated, its method has expanded to the coloured image copyright protection, also just simply gray level image has been extended to coloured image, has not significantly upgraded on the algorithm; And the method in the coloured image copyright protection need become YCbCr with the RGB mode switch of coloured image.

Summary of the invention

The purpose of this invention is to provide a kind of based on the secret picture material altering detecting method of sharing of vision, utilize the copyright information of image to generate watermarking images, do not need watermark embedding and watermark extracting, but utilize the secret technology of sharing of vision, raw image data is not being produced under the prerequisite of any destruction, realizing the detection of distorting simultaneously picture material.

The technical solution adopted in the present invention is that a kind of picture material altering detecting method of sharing based on the vision secret may further comprise the steps:

Information sender:

Step 1, for given original image I ₀, extract original image feature F with feature extraction algorithm ₀

Step 2, utilize original image I ₀Copyright information generate original watermark image W ₀, encrypt original watermark image W with cryptographic algorithm again ₀, generate encrypted watermark image W _E

Step 3, the original image feature F that utilizes step 1 to obtain ₀And the encrypted watermark image W that obtains of step 2 _E, generate checking shadow S by cryptographic algorithm Algorithm 3 ₁

The receiving party:

Step 4, for the test pattern I that receives _t, extract detected image feature F with feature extraction algorithm _t

Step 5, the detected image feature F that utilizes step 4 to obtain _tAnd the encrypted watermark image W that obtains of step 2 _E, generate detection shadow S by algorithm Algorithm 4 ₂

Step 6, the checking shadow S that utilizes step 3 to obtain ₁And the detection shadow S that obtains of step 5 ₂,, generate recovery watermarking images W by watermark recovery algorithm Algorithm 5 _R

Step 7, the recovery watermarking images W that utilizes decipherment algorithm decryption step 6 to obtain _R, generate and detect watermarking images W _D

Step 8, the detection watermarking images W that step 7 is obtained _DThe original watermark image W that obtains with step 2 ₀Compare, if detect watermarking images W _DWith original watermark image W ₀Visual effect is identical, then shows test pattern I _tBe by original image I ₀Through obtaining after the normal image processing operational processes; If detect watermarking images W _DSerious distortion then shows and detects watermarking images W _DBe the image of malice after distorting, and according to watermarking images W _DTest pattern I is determined in the position of distortion _tBy the tampered region.

The invention has the beneficial effects as follows the following deficiency that has solved existing correlation technique:

(1) solved deficiency, do not needed watermarking algorithm and watermark extracting algorithm based on the image authentication method of digital watermarking, at the bottom of the computation complexity, the efficient height.

(2) raw image data is not being produced under the prerequisite of any destruction, realizing the detection of distorting simultaneously, to realize copyright protection to picture material.

(3) the present invention can directly apply to coloured image, does not need to carry out the conversion of color mode.

(4) method of the present invention for JPEG compression, image sharpening, interpolation noise, slightly rotate normal image such as convergent-divergent and handle operation and have robustness.

(5) malice is distorted have susceptibility, can determine by the position of tampered region.

(6), thereby has stronger practicality because method of the present invention has higher security and efficient.

Description of drawings

Fig. 1 is the process flow diagram that the present invention is based on the secret picture material altering detecting method of sharing of vision;

Fig. 2 is that quality factor is the image that obtains in the content tampering testing process of 20 jpeg compressed image, and wherein, a is original image I ₀, b is original watermark image W ₀, c is encrypted watermark image W _E, d is checking shadow S ₁, e is for detecting shadow S ₂, f is for recovering watermarking images W _R, g is test pattern I _t, h is for detecting watermarking images W _D

Fig. 3 is that quality factor is the image that obtains in the content tampering testing process of 60 jpeg compressed image, and wherein, a is original image I ₀, b is original watermark image W ₀, c is encrypted watermark image W _E, d is checking shadow S ₁, e is for detecting shadow S ₂, f is for recovering watermarking images W _R, g is test pattern I _t, h is for detecting watermarking images W _D

Fig. 4 is the content tampering detection figure at gray level image, and a is test pattern I _t, b is the tampering location result;

Fig. 5 is the content tampering detection figure at coloured image, and a is test pattern I _t, b is the tampering location result.

Embodiment

The present invention is described in detail below in conjunction with the drawings and specific embodiments.

As shown in Figure 1, the present invention is a kind of based on the secret picture material altering detecting method of sharing of vision, may further comprise the steps:

Information sender:

Step 1, for given original image I ₀, extract original image feature F with feature extraction algorithm Algorithm 1 ₀

Wherein, the feature extraction algorithm Algorithm in the step 11 is:

The original image I of the N * N of input ₀Be divided into 8 * 8 not overlapping block, obtain N ²/ 64 image blocks are expressed as I _Ij, i=1 wherein, 2 ..., N/8, j=1,2 ..., N/8; Each image block is carried out 3 grades of db1 wavelet decomposition, obtain the LL of each 8 * 8 image block ₃Subband wavelet coefficient a _Ij(i=1,2 ..., N/8, j=1,2 ..., N/8), generate original image feature F ₀:

$F_0=\left[\begin{array}{ccc}{a}_{11}& ...& a_{1n}\\ ...& ...& ...\\ a_{\mathrm{<figure-callout\; id="1"\; label="n"\; filenames="HDA0000049451620000011.png"\; state="\{\{state\}\}">n</figure-callout>}1}& ...& a_{\mathrm{nn}}\end{array}\right],$ N=N/8 wherein.

Step 2, utilize original image I ₀Copyright information generate original watermark image W ₀, encrypt original watermark image W with cryptographic algorithm Algorithm 2 again ₀, generate encrypted watermark image W _E

Cryptographic algorithm Algorithm 2 in the step 2 is:

Wherein, utilize original image I ₀Copyright information generate the original watermark image W of N/8 * N/8 ₀, construct a N/8-rank magic square matrix, through K (K＜(N/8) ²) after the magic square conversion, obtain encrypted watermark image W _EWherein, K is given watermark keys, encrypted watermark image W _EBe a binary matrix:

$W_E=\left[\begin{array}{ccc}{w}_{11}& ...& w_{1n}\\ ...& ...& ...\\ w_{n1}& ...& w_{\mathrm{nn}}\end{array}\right],$ N=N/8 wherein, w _Ij∈ (0,1).

Step 3, the original image feature F that utilizes step 1 to obtain ₀And the encrypted watermark image W that obtains of step 2 _E,, generate checking shadow S by cryptographic algorithm Algorithm 3 ₁

Wherein, the cryptographic algorithm Algorithm in the step 33 is:

At first, the characteristics of image F that obtains of calculation procedure 1 ₀Average M ₀: $M_0=\frac{a_{11}+...+a_{1(N/8)}+...+a_{(N/8)1}+...+a_{(N/8)(N/8)}}{(N/8)\&times;(N/8)};$

Secondly, ${\mathrm{<figure-callout\; id="1"\; label="S"\; filenames="HDA0000049451620000011.png"\; state="\{\{state\}\}">S</figure-callout>}}_1=\left[\begin{array}{ccc}{s}_{11}& ...& s_{1n}\\ ...& ...& ...\\ s_{n1}& ...& s_{\mathrm{nn}}\end{array}\right]=F_0{e}_1{W}_E=\left[\begin{array}{ccc}{a}_{11}{e}_1{w}_{11}& ...& a_{1n}{e}_1{w}_{1n}\\ ...& ...& ...\\ a_{n1}{e}_1{\mathrm{<figure-callout\; id="1"\; label="w\; n"\; filenames="HDA0000049451620000011.png"\; state="\{\{state\}\}">w</figure-callout>}}_n& ...& a_{\mathrm{nn}}{e}_1{w}_{\mathrm{nn}}\end{array}\right]$

Wherein, n=N/8, algorithm " e ₁" definition rule be:

	w ij＝1	w ij＝0
			a ij＜M 0	s ij＝1	s ij＝0
a ij＞M 0	s ij＝0	s ij＝1

The receiving party:

Step 4, for the test pattern I that receives _t, extract detected image feature F with feature extraction algorithm Algorithm 1 _t

Wherein, the algorithm Algorithm in the step 41 is:

The test pattern I of the N * N that receives _tBe divided into 8 * 8 not overlapping block, obtain N ²/ 64 image blocks carry out 3 grades of db1 wavelet decomposition to each image block, obtain the LL of each 8 * 8 image block ₃Subband wavelet coefficient b _Ij(i=1,2 ..., N/8, j=1,2 ..., N/8), generate detected image feature F _t:

$F_t=\left[\begin{array}{ccc}{b}_{11}& ...& b_{1n}\\ ...& ...& ...\\ b_{n1}& ...& b_{\mathrm{nn}}\end{array}\right],$ N=N/8 wherein.

Step 5, the detected image feature F that utilizes step 4 to obtain _tAnd the encrypted watermark image W that obtains of step 2 _E,, generate detection shadow S by algorithm Algorithm 4 ₂

Wherein, the algorithm Algorithm in the step 54 is:

At first, the detected image feature F that obtains of calculation procedure 4 _tAverage M _t: $M_t=\frac{b_{11}+...+b_{1(N/8)}+...+b_{(N/8)1}+...+b_{(N/8)(N/8)}}{(N/8)\&times;(N/8)};$

Secondly, $S_2=\left[\begin{array}{ccc}{t}_{11}& ...& t_{1n}\\ ...& ...& ...\\ t_{n1}& ...& t_{\mathrm{nn}}\end{array}\right]=F_t{e}_2{W}_E=\left[\begin{array}{ccc}{b}_{11}{e}_2{w}_{11}& ...& b_{1n}{e}_2{w}_{1n}\\ ...& ...& ...\\ b_{n1}{e}_2{\mathrm{<figure-callout\; id="1"\; label="w\; n"\; filenames="HDA0000049451620000011.png"\; state="\{\{state\}\}">w</figure-callout>}}_n& ...& b_{\mathrm{nn}}{e}_2{w}_{\mathrm{nn}}\end{array}\right],$

Wherein, n=N/8, algorithm " e ₂" definition rule be:

	w ij＝1	w ij＝0
			b ij＜M t	t ij＝0	t ij＝0
b ij＞M t	t ij＝1	t ij＝1

Step 6, the checking shadow S that utilizes step 3 to obtain ₁And the detection shadow S that obtains of step 5 ₂,, generate recovery watermarking images W by watermark recovery algorithm Algorithm 5 _R

Wherein, the watermark recovery algorithm Algorithm 5 in the step 6 is:

$W_R=\left[\begin{array}{ccc}{w}_{R11}& ...& w_{R1n}\\ ...& ...& ...\\ w_{\mathrm{<figure-callout\; id="1"\; label="Rn"\; filenames="HDA0000049451620000011.png"\; state="\{\{state\}\}">Rn</figure-callout>}1}& ...& w_{\mathrm{Rnn}}\end{array}\right]=S_1{e}_3{S}_2=\left[\begin{array}{ccc}{s}_{11}{e}_3{t}_{11}& ...& s_{1n}{e}_3{t}_{1n}\\ ...& ...& ...\\ s_{n1}{e}_3{\mathrm{<figure-callout\; id="1"\; label="t\; n"\; filenames="HDA0000049451620000011.png"\; state="\{\{state\}\}">t</figure-callout>}}_n& ...& s_{\mathrm{nn}}{e}_3{t}_{\mathrm{nn}}\end{array}\right],$

Wherein, n=N/8, algorithm " e ₃" definition rule be:

s ij	t ij	s ij?e 3?t ij
			0	1	1
1	0	1
			1	1	0
0	0	0

Step 7, the recovery watermarking images W that utilizes decipherment algorithm Algorithm 6 decryption step 6 to obtain _R, generate and detect watermarking images W _D

Wherein, the decipherment algorithm Algorithm in the step 76 is: utilize watermarking images W _RConstruct a N/8-rank magic square matrix, carry out K contrary magic square conversion.

Step 8, the detection watermarking images W that step 7 is obtained _DThe original watermark image W that obtains with step 2 ₀Compare: if detect watermarking images W _DWith original watermark image W ₀Visual effect is identical, then shows test pattern I _tBe by original image I ₀Through obtaining after the normal image processing operational processes; If detect watermarking images W _DSerious distortion then shows and detects watermarking images W _DBe the image of malice after distorting, and according to watermarking images W _DTest pattern I is determined in the position of distortion _tBy the tampered region.

As shown in the table is the comparison of method of the present invention and existing technologies:

In last table, the function that " √ " expression algorithm has; The function that " * " expression algorithm does not have.As can be seen, method of the present invention also has the positioning function in the field of distorting except the function that has existing technologies and have.

As shown in Figure 2, be that quality factor is the image that obtains in the content tampering testing process of 20 jpeg compressed image.Fig. 3 is that quality factor is the image that obtains in the content tampering testing process of 60 jpeg compressed image, and in Fig. 2 and Fig. 3, a is given original image I ₀B is the original watermark image W that step 2 generates ₀C is the encrypted watermark image W that step 2 generates _ED is the checking shadow S that step 3 generates ₁E is the detection shadow S that step 5 generates ₂F is the recovery watermarking images W that step 6 generates _RG is test pattern I _t, h is the detection watermarking images W that step 7 generates _DThe picture material that draws Fig. 2 and Fig. 3 by comparison b and h is distorted test result and is: test pattern I _tBe by original image I ₀Through obtaining after the normal image processing operational processes.

Fig. 4 is the content tampering detection figure at gray level image, and a is test pattern I _t, b is the tampering location result.Fig. 5 is the content tampering detection figure at coloured image, and a is test pattern I _t, b is the tampering location result.

Claims (5)

1. one kind based on the secret picture material altering detecting method of sharing of vision, it is characterized in that may further comprise the steps:

Information sender:

Step 1, for given original image I ₀, extract original image feature F with feature extraction algorithm ₀

Step 2, utilize original image I ₀Copyright information generate original watermark image W ₀, encrypt original watermark image W with cryptographic algorithm again ₀, generate encrypted watermark image W _E

Step 3, the original image feature F that utilizes step 1 to obtain ₀And the encrypted watermark image W that obtains of step 2 _E, generate checking shadow S by cryptographic algorithm Algorithm 3 ₁, wherein, described cryptographic algorithm Algorithm 3 is:

At first, the characteristics of image F that obtains of calculation procedure 1 ₀Average M ₀: $M_0=\frac{a_{11}+...+a_{1(N/8)}+...+a_{(N/8)1}+...+a_{(N/8)(N/8)}}{(N/8)\&times;(N/8)};$

Secondly, $S_1=\left[\begin{array}{ccc}{s}_{11}& ...& s_{1n}\\ ...& ...& ...\\ s_{n1}& ...& s_{\mathrm{nn}}\end{array}\right]=F_0{e}_1{W}_E=\left[\begin{array}{ccc}{a}_{11}{e}_1{w}_{11}& ...& a_{1n}{e}_1{w}_{1n}\\ ...& ...& ...\\ a_{n1}{e}_1{w}_{n1}& ...& a_{\mathrm{nn}}{e}_1{w}_{\mathrm{nn}}\end{array}\right]$

Wherein, n=N/8, algorithm " e ₁" definition rule be:

w _ij＝1 w _ij＝0 a _ij＜M ₀ s _ij＝1 s _ij＝0 a _ij＞M ₀ s _ij＝0 s _ij＝1

；

The receiving party:

Step 4, for the test pattern I that receives _t, extract detected image feature F with feature extraction algorithm _t

Step 5, the detected image feature F that utilizes step 4 to obtain _tAnd the encrypted watermark image W that obtains of step 2 _E, generate detection shadow S by algorithm Algorithm 4 ₂, wherein, described algorithm Algorithm4 is:

At first, the detected image feature F that obtains of calculation procedure 4 _tAverage M _t: $M_t=\frac{b_{11}+...+b_{1(N/8)}+...+b_{(N/8)1}+...+b_{(N/8)(N/8)}}{(N/8)\&times;(N/8)};$

Secondly, $S_2=\left[\begin{array}{ccc}{t}_{11}& ...& t_{1n}\\ ...& ...& ...\\ t_{n1}& ...& t_{\mathrm{nn}}\end{array}\right]=F_t{e}_2{W}_E=\left[\begin{array}{ccc}{b}_{11}{e}_2{w}_{11}& ...& b_{1n}{e}_2{w}_{1n}\\ ...& ...& ...\\ b_{n1}{e}_2{w}_{n1}& ...& b_{\mathrm{nn}}{e}_2{w}_{\mathrm{nn}}\end{array}\right],$

Wherein, n=N/8, algorithm " e ₂" definition rule be:

w _ij＝1 w _ij＝0 b _ij＜M _t t _ij＝0 t _ij＝0

b _ij＞M _t t _ij＝1 t _ij＝1

；

Step 6, the checking shadow S that utilizes step 3 to obtain ₁And the detection shadow S that obtains of step 5 ₂,, generate recovery watermarking images W by watermark recovery algorithm Algorithm 5 _R, wherein, described step watermark recovery algorithm Algorithm 5 is:

$W_R=\left[\begin{array}{ccc}{w}_{R11}& ...& w_{R1n}\\ ...& ...& ...\\ w_{\mathrm{Rn}1}& ...& w_{\mathrm{Rnn}}\end{array}\right]=S_1{e}_3{S}_2=\left[\begin{array}{ccc}{s}_{11}{e}_3{t}_{11}& ...& s_{1n}{e}_3{t}_{1n}\\ ...& ...& ...\\ s_{n1}{e}_3{t}_{n1}& ...& s_{\mathrm{nn}}{e}_3{t}_{\mathrm{nn}}\end{array}\right],$

Wherein, n=N/8, algorithm " e ₃" definition rule be:

s _ij t _ij s _ij□ ₃t _ij 0 1 1 1 0 1 1 1 0 0 0 0

；

Step 7, the recovery watermarking images W that utilizes decipherment algorithm decryption step 6 to obtain _R, generate and detect watermarking images W _D

Step 8, the detection watermarking images W that step 7 is obtained _DThe original watermark image W that obtains with step 2 ₀Compare, if detect watermarking images W _DWith original watermark image W ₀Visual effect is identical, then shows test pattern I _tBe by original image I ₀Through obtaining after the normal image processing operational processes; If detect watermarking images W _DSerious distortion then shows and detects watermarking images W _DBe the image of malice after distorting, and according to watermarking images W _DTest pattern I is determined in the position of distortion _tBy the tampered region.

2. described based on the secret picture material altering detecting method of sharing of vision according to claim 1, it is characterized in that the feature extraction algorithm in the described step 1 is:

The original image I of the N * N of input ⁰Be divided into 8 * 8 not overlapping block, obtain N ²/ 64 image blocks are expressed as I ^Ij, i=1 wherein, 2 ..., N/8, j=1,2 ..., N/8; Each image block is carried out 3 grades of db1 wavelet decomposition, obtain the LL of each 8 * 8 image block ₃Subband wavelet coefficient a _Ij(i=1,2 ..., N/8, j=1,2 ..., N/8), generate original image feature F ₀:

$F_0=\left[\begin{array}{ccc}{a}_{11}& ...& a_{1n}\\ ...& ...& ...\\ a_{n1}& ...& a_{\mathrm{nn}}\end{array}\right],$ N=N/8 wherein.

3. described based on the secret picture material altering detecting method of sharing of vision according to claim 1, it is characterized in that the cryptographic algorithm in the described step 2 is:

Utilize original image I ₀Copyright information generate the original watermark image W of N/8 * N/8 ₀, construct a N/8-rank magic square matrix, through after K the magic square conversion, obtain encrypted watermark image W _EWherein, K is given watermark keys, K＜(N/8) ², encrypted watermark image W _EBe a binary matrix:

$W_E=\left[\begin{array}{ccc}{w}_{11}& ...& w_{1n}\\ ...& ...& ...\\ w_{n1}& ...& w_{\mathrm{nn}}\end{array}\right],$ N=N/8 wherein, w _Ij∈ (0,1).

4. described based on the secret picture material altering detecting method of sharing of vision according to claim 1, it is characterized in that the feature extraction algorithm in the described step 4 is:

The test pattern I of the N * N that receives _tBe divided into 8 * 8 not overlapping block, obtain N ²/ 64 image blocks carry out 3 grades of db1 wavelet decomposition to each image block, obtain the LL of each 8 * 8 image block ₃Subband wavelet coefficient b _Ij, wherein, i=1,2 ..., N/8, j=1,2 ..., N/8 generates detected image feature F _t:

$F_t=\left[\begin{array}{ccc}{b}_{11}& ...& b_{1n}\\ ...& ...& ...\\ b_{n1}& ...& b_{\mathrm{nn}}\end{array}\right],$ N=N/8 wherein.

5. described based on the secret picture material altering detecting method of sharing of vision according to claim 1, it is characterized in that the decipherment algorithm in the described step 7 is: utilize watermarking images W _RConstruct a N/8-rank magic square matrix, carry out K contrary magic square conversion, wherein, K is given watermark keys, K＜(N/8) ²

Images