CN101493939A - Method for detecting cooked image based on small wave domain homomorphic filtering - Google Patents

Abstract

The invention discloses a method for detecting a forgery image on the basis of homomorphic filtering in the wavelet domain; after HIS transformation is carried out to a colored image, single-layer wavelet decomposition is carried out to the brightness component of the colored image; a homomorphic filtering function in the wavelet domain is designed; a filter coefficient is utilized to carry out homomorphic filtering in the wavelet domain to the image after single-layer wavelet decomposition; similar high pass filtering processing is carried out to wavelet decomposition coefficients of different resolution; an edge detection operator is utilized to extract the information of an edge image after homomorphic filtering in the wavelet domain; a structural element is selected to carry out corrosion operation to the information of the edge image; an edge after being manually blurred is extracted; a connected region of the image after corrosion is marked; a threshold value is set according to the results of experiments; and the connected region with an area not less than 0.8 percent of the connected region of the original image is retained, which is the forgery area of the image. The method effectively blurred edge, thereby lowering the false drop rate.

Description

Forge the method for image based on the detection of wavelet field homomorphic filtering

Technical field

The present invention relates to the image content information security fields, particularly picture material authenticity blind checking method.

Background technology

Current protection to digital image content mainly utilizes digital watermarking and signature technology to realize, the basic thought that these technology adopt all is by adding additional information image to be carried out the authenticity discriminating.But do not contain digital watermarking or signature in present most image.Along with popularizing of digital camera and image processing software with popular, increasing high-quality forgery image is propagated in a large number by network, " soon " that makes people be " for real " no longer, caused the negative effect of aspects such as politics, culture, news and science authenticity.Further, a large amount of existence of forging photo have influence on individual's consciousness of the public probably, finally allow people lose the trust of comparison film.Therefore, carrying out of this research has military significance, civilian meaning and scientific meaning to society and people's life.

It is one of branch of information security technology that the digital picture authenticity detects, and is the technology of under the situation of sign such as embed watermark the authenticity of digital image content being estimated in image in advance not.This technology is come that image is carried out the true and false and is detected by analysis of image data and statistical property thereof, and to the forgery in the image with alter vestige and detect and locate.

At present, the research of content reliability of digital picture evaluation still is in the starting stage,, has obtained some achievements in research along with domestic and international associated mechanisms improves day by day to its degree of concern.These researchs are primarily aimed at certain and specifically distort operation and image is carried out the true and false detect, comprising fuzzy operation, and dual JPEG squeeze operation, partial copy-paste operation, re-sampling operations, concatenation or the like.These technology all are based on such hypothesis: exist some statistical character in natural image, then can change its potential statistical law if view data made amendment.This hypothesis also is to judge whether piece image is modified and carries out the foundation of tampering location.

Aspect synthetic edge, occurred some and detected the technology of the artificial fuzzy operation vestige of synthetic edge.Patent publication No. is CN1862598 (open day: in image forge 2006.11.15) in the method for obtaining evidence of abnormal hue rate of fuzzy operation, utilize fuzzy operation to digital picture local color attribute cause unusual, by set of definition abnormal hue and abnormal hue rate, locally coherence and degree of relevancy to picture tone quantize, and then the digital picture of a width of cloth through forgery and artificial fuzzy operation detected.People such as Zhou Linna publish an article in the KES-AMSTA2007 meeting, utilize the method for homomorphic filtering, moving average filtering and mathematical morphology that the digital picture of crossing through artificial Fuzzy Processing is distorted detection.But with respect to performance image editing software from strength to strength and the adulterator who is skilled in technique, these methods and achievement in research still strength a little less than, also can't fundamentally contain the paces of image forge.

Summary of the invention

In order to overcome the effectively deficiency of the detected image true and false of prior art, the invention provides a kind of method of forging image based on the detection of wavelet field homomorphic filtering, summed up the weak point rule that may exist the forgery image from synthetic edge feature aspect, analyzed the vestige that the image forge process may be left over, and accurately located forging the zone.

Principle analysis of the present invention is as follows:

At present, image is synthetic to be a kind of method of very common forgery image, most of images through simple synthetic after, in order to eliminate vision or the statistical distortion that produces at synthetic edge, all can adopt operations such as fuzzy, emergence, gradual change to carry out post-processed.No matter adopt which kind of retouching operation, all inevitably to influence below the synthetic edge generation of image: after various retouching operations, artificial fuzzy edge grey level range reduces, be hard to tell the gray-level and the details of object, and all the other general grayscale dynamic range in undressed normal edge are bigger, the level distinctness.So, the edge after the Fuzzy Processing is strengthened amplifying, thereby make fuzzy edge have the characteristic of broad, utilize the morphological erosion operation that relative more weak natural edge corrosion is fallen then, the fuzzy edge that stays enhancing is forged the zone with the location.

Fuzzy edge is strengthened amplifying, adopt the general linear conversion to be difficult to prove effective, though, can make dynamic range bigger because unified expanded images gray level can improve the contrast of subject image, also strengthen simultaneously the normal edge of unprocessed part, caused erroneous judgement easily; The gray level of unified compressed image makes the gray-level and the details of artificial Fuzzy Processing part more unclear, then can't setting threshold when evidence obtaining.The wavelet field homographic filtering method, coefficient of wavelet decomposition under the different resolution is carried out similar high-pass filtering to be handled, can be under the situation that does not influence the nature edge, strengthened the artificial fuzzy part edge of corresponding high band reflecting component well, the morphological erosion operation can be fallen relative more weak natural edge corrosion, then the fuzzy edge that carries over is adopted the region labeling method, orient the image forge zone.

The analysis of tradition homomorphism is from the integral image angle image to be handled, the original looks that can keep image well, but it does not take into full account the spatial domain local feature of image, the enhancing effect of high frequency composition still can not be satisfactory, and fuzzy part is corresponding to the high band reflecting component after the neighborhood averaging gray scale, so that traditional homomorphic filtering strengthens the effect of artificial fuzzy edge is undesirable.

The technical solution adopted for the present invention to solve the technical problems may further comprise the steps:

(1) coloured image is carried out the HIS conversion, obtain brightness, colourity and three components of color saturation, the formula of HIS conversion is as follows:

$\left\{\begin{array}{c}I=\frac{R+G+B}{3}\\ H=\frac{1}{360}[90-\arctan \left(\frac{2R-G-B}{\sqrt{3}(G-B)}\right)+\{0,G>B;180,G<B\left\}\right]\\ S=1-\left[\frac{\min (R,G,B)}{I}\right]\end{array}\right.---\left(1\right)$

Wherein I is brightness, and H is a colourity, and S is a color saturation, and R represents the red component of coloured image, and G represents the green component of coloured image, and B represents the blue component of coloured image.

(2) luminance component to coloured image carries out one deck wavelet decomposition, obtains each the different components behind the wavelet transformation.

Wavelet transformation is that original signal is decomposed into a plurality of high frequency band data and low-frequency band data on a series of frequencys multiplication.Image through J layer wavelet decomposition, is obtained LH _j, HL _j, HH _j(j=1,2 ..., J) and LL _J, LH wherein _j, HL _j, HH _jRepresent horizontal direction after the j layer wavelet decomposition, vertical direction respectively and to the high fdrequency component of angular direction, LL _JRepresent the low frequency component after the J layer wavelet decomposition.

(3) design wavelet field homomorphic filtering function.

The expression formula of exponential homomorphism filter function commonly used is as follows in the reality.

$H(u,v)=({\mathrm{\&gamma;}}_H-{\mathrm{\&gamma;}}_L)[1-e^{-c{\left(\frac{D(u,v)}{D_0}\right)}^2}]+{\mathrm{\&gamma;}}_L---\left(2\right)$

D wherein ₀Be cutoff frequency, $D(u,v)=\sqrt{u^2+v^2},$ U, v are frequency variable, and constant c is introduced into and is used for the sharpening on control filters function inclined-plane, and it is at γ _HAnd γ _LBetween transition, and γ _H＞1, γ _L＜1, γ _HAnd γ _LBe two parameters that are used for control characteristic homomorphic filter shape.

The exponential homomorphism filter function is improved, so that it can be used for the Homomorphic Filtering Algorithm of wavelet field, the filter function after the improvement is as follows.

$H(j,w_h,w_v)=({\mathrm{\&gamma;}}_H-{\mathrm{\&gamma;}}_L)[1-e^{-{\left(\frac{\sqrt{w_h^2+w_v^2}}{2^j{k}_c}\right)}^2}]+{\mathrm{\&gamma;}}_L---\left(3\right)$

Wherein j represents the wavelet decomposition number of plies, k _cFor ending coefficient (0＜k _c＜1), w _hAnd w _vBe respectively horizontal weights coefficient and vertical weights coefficient.For LH _j, w _h=0, w _v=1; For HL _j, w _h=1, w _c=0; For HH _j, w _h=1, w _v=1.The frequency domain frequency multiplication feature of wavelet coefficients at different levels is satisfied in this improvement, has guaranteed the relation between filtering characteristic and filtering parameter in traditional Hi-pass filter simultaneously.

For LL _JOn coefficient can carry out the irregularity in brightness that image is adjusted in linear or nonlinear adjustment according to the characteristics of image.Take linear adjustment mode herein, as the formula (4).

$H_{{\mathrm{LL}}_J}={\mathrm{\&gamma;}}_L(k(x-m)+m)---\left(4\right)$

Here, expression formula k (x-m)+m finishes LL _JThe linear equalization of coefficient, x represents LL _JOn wavelet coefficient, m represents LL _JThe mean value of last wavelet coefficient, contrast adjustment factor k satisfies 0≤k≤1.

(4) image after utilizing the filter coefficient that obtains in the step (3) to one deck wavelet decomposition carries out the wavelet field homomorphic filtering, the coefficient of wavelet decomposition under the different resolution is carried out similar high-pass filtering handle, and with the decay low-frequency component, strengthens high-frequency information.The process of wavelet field homomorphic filtering such as accompanying drawing 1 show, wherein HL, LH, HH and four different components after LL represents wavelet transformation, and H_HL, H_LH, H_HH and H_LL represent the filter coefficient of each component.

(5) utilize edge detection operator (as Sobel operator, Laplace operator etc.) to extract edge of image image information after the wavelet field homomorphic filtering, the Sobel operator that experiment is adopted in the invention is as follows:

$H_1=\left[\begin{array}{ccc}-1& -2& -1\\ 0& 0& 0\\ 1& 2& 1\end{array}\right],$ $H_2=\left[\begin{array}{ccc}-1& 0& 1\\ -2& 0& 2\\ -1& 0& 1\end{array}\right]$

Wherein, H ₁Be used for extracting horizontal edge, H ₂Be used for extracting vertical edge.Marginal information herein is that horizontal edge and vertical edge that extraction obtains are averaged processing.

(6) the artificial blur radius that blurs mostly is 3～7 pixels greatly, so select 3 * 3 or 5 * 5 structural element SE that the edge image information is carried out erosion operation, corrosion shrinks without the normal edge of image that strengthens, and extracts the edge after artificial bluring.

(7) connected region of image after the employing eight connected region standardization indicia etched according to the experimental result setting threshold, keeps the connected region that area is not less than original image 0.8%, and this connected region is the image forge zone.

The invention has the beneficial effects as follows: owing to adopt the wavelet field homographic filtering method, coefficient of wavelet decomposition under the different resolution is carried out similar high-pass filtering to be handled, strengthened fuzzy region " wide " edge, combining form is learned the corrosion operation, image nature edge corrosion is fallen, thereby reservation is oriented through the artificial fuzzy edge that strengthens and is forged the zone, with compare based on traditional homographic filtering method, owing to realize the gray scale dynamic range of compression entire image preferably, make fuzzy edge gray level expansion simultaneously, effectively strengthen fuzzy edge, thereby reduced false drop rate.

The present invention is further described below in conjunction with drawings and Examples.

Description of drawings

Fig. 1 is a wavelet field homomorphic filtering process synoptic diagram.

Fig. 2 and Fig. 3 utilize the wavelet field homomorphic filtering to detect sample image and the testing result synoptic diagram of forging image.

Embodiment

(a) and (b) are original image among Fig. 2, and (c) for forging image, it is the aircraft among the former figure (b) to be copied to obtain in (a), and have used post-processing operation such as edge fog.Utilizing the wavelet field homomorphic filtering that it is forged image (c) detects.(d) being depicted as the wavelet transformation of forging luminance component image, (e) is the image after the wavelet field homomorphic filtering, (f) is testing result.

(a) and (b) are original image among Fig. 3, and (c) for forging image, it is the mouse among the former figure (b) to be copied to obtain in (a), and have used post-processing operation such as edge fog.We utilize the wavelet field homomorphic filtering that it is forged image (c) and detect.(d) being depicted as the wavelet transformation of forging luminance component image, (e) is the image after the wavelet field homomorphic filtering, (f) is testing result.The detection step is as follows:

(1) coloured image in the experiment (Fig. 2 (c) and Fig. 3 (c)) is carried out the HIS conversion, obtain brightness, colourity and three components of color saturation;

(2) select Biro4.4 biorthogonal wavelet base that the luminance component of the coloured image that obtains is carried out one deck wavelet decomposition; (Fig. 2 (d) and Fig. 3 (d));

(3), obtain filter coefficient H_HL, H_LH, H_HH and the H_LL of each component behind the wavelet transformation according to formula (3) and (4) design wavelet field homomorphic filtering function;

(4) image after utilizing the filter coefficient that obtains in the step (3) to one deck wavelet decomposition carries out the wavelet field homomorphic filtering, obtain the image (Fig. 2 (e) and Fig. 3 (e)) after the wavelet field homomorphic filtering, wherein index homomorphic filter function parameters is provided with as follows among Fig. 2: γ _H=2.2, γ _L=0.9, $k_c=\frac{1}{8},$ K=0.8, the parameter among Fig. 3 is provided with as follows: γ _H=3.1, γ _L=1.3, $k_c=\frac{1}{6},$ k＝0.66；

(5) utilize the Sobel operator ( $H_1=\left[\begin{array}{ccc}-1& -2& -1\\ 0& 0& 0\\ 1& 2& 1\end{array}\right],$ $H_2=\left[\begin{array}{ccc}-1& 0& 1\\ -2& 0& 2\\ -1& 0& 1\end{array}\right]$ ) extraction edge of image image information after the wavelet field homomorphic filtering;

(6) select 3 * 3 structural elements $\mathrm{SE}=\left[\begin{array}{ccc}1& 1& 1\\ 1& 1& 1\\ 1& 1& 1\end{array}\right]$ Edge image is carried out erosion operation, and corrosion shrinks without the normal edge of image that strengthens, and extracts the edge after artificial bluring;

(7) connected region of image after the employing eight connected region standardization indicia etched according to the experimental result setting threshold, keeps the connected region that area is not less than original image 0.8%, and this connected region is the image forge zone.(Fig. 2 (f) and Fig. 3 (f)).

Claims (2)

1, forges the method for image based on the detection of wavelet field homomorphic filtering, it is characterized in that comprising the steps:

(1) coloured image is carried out the HIS conversion, formula is as follows:

$\left\{\begin{array}{c}I=\frac{R+G+B}{3}\\ H=\frac{1}{360}[90-\arctan \left(\frac{2R-G-B}{\sqrt{3}(G-B)}\right)+\{0,G>B;180,G<B\left\}\right]\\ S=1-\left[\frac{\min (R,G,B)}{I}\right]\end{array}\right.$

Wherein I is brightness, and H is a colourity, and S is a color saturation, and R represents the red component of coloured image, and G represents the green component of coloured image, and B represents the blue component of coloured image;

(2) luminance component to coloured image carries out one deck wavelet decomposition, obtains each the different component LH behind the wavelet transformation _j, HL _j, HH _j(j=1,2 ..., J) and LL _J, LH wherein _j, HL _j, HH _jRepresent horizontal direction after the j layer wavelet decomposition, vertical direction respectively and to the high fdrequency component of angular direction, LL _JRepresent the low frequency component after the J layer wavelet decomposition;

(3) design wavelet field homomorphic filtering function $H(j,w_h,w_v)=({\mathrm{\&gamma;}}_H-{\mathrm{\&gamma;}}_L)[1-e^{-{\left(\frac{\sqrt{w_h^2+w_v^2}}{2^j{k}_c}\right)}^2}]+{\mathrm{\&gamma;}}_L,$ Wherein j represents the wavelet decomposition number of plies, k _cFor ending coefficient (0＜k _c＜1), w _hAnd w _vBe respectively horizontal weights coefficient and vertical weights coefficient, for LH _j, w _h=0, w _v=1; For HL _j, w _h=1, w _v=0; For HH _j, w _h=1, w _v=1;

For LL _J, HLL _J=γ _L(k (x-m)+m), wherein, x represents LL _JOn wavelet coefficient, m represents LL _JThe mean value of last wavelet coefficient, contrast adjustment factor k satisfies 0≤k≤1;

(4) image after utilizing the filter coefficient that obtains in the step (3) to one deck wavelet decomposition carries out the wavelet field homomorphic filtering, the coefficient of wavelet decomposition under the different resolution is carried out similar high-pass filtering handle, and with the decay low-frequency component, strengthens high-frequency information;

(5) utilize edge detection operator to extract edge of image image information after the wavelet field homomorphic filtering;

(6) the structural element SE of selection 3 * 3 or 5 * 5 carries out erosion operation to the edge image information, and corrosion shrinks without the normal edge of image that strengthens, and extracts the edge after artificial bluring;

(7) connected region of image after the employing eight connected region standardization indicia etched according to the experimental result setting threshold, keeps the connected region that area is not less than original image 0.8%, and this connected region is the image forge zone.

2, method of forging image based on the detection of wavelet field homomorphic filtering according to claim 1 is characterized in that: the Sobel operator that described edge detection operator adopts:

$H_1=\left[\begin{array}{ccc}-1& -2& -1\\ 0& 0& 0\\ 1& 2& 1\end{array}\right],$ $H_2=\left[\begin{array}{ccc}-1& 0& 1\\ -2& 0& 2\\ -1& 0& 1\end{array}\right]$

Wherein, H ₁Be used for extracting horizontal edge, H ₂Be used for extracting vertical edge.

Images