CN103208100B - Blurred kernel inversion method for blurred retouching images based on blurred deformation Riemann measure - Google Patents

Blurred kernel inversion method for blurred retouching images based on blurred deformation Riemann measure Download PDF

Info

Publication number
CN103208100B
CN103208100B CN201310089605.8A CN201310089605A CN103208100B CN 103208100 B CN103208100 B CN 103208100B CN 201310089605 A CN201310089605 A CN 201310089605A CN 103208100 B CN103208100 B CN 103208100B
Authority
CN
China
Prior art keywords
blurred
fuzzy
image
retouching
riemann
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201310089605.8A
Other languages
Chinese (zh)
Other versions
CN103208100A (en
Inventor
杜振龙
金雨霏
李晓丽
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Nanjing Tech University
Original Assignee
Nanjing Tech University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Nanjing Tech University filed Critical Nanjing Tech University
Priority to CN201310089605.8A priority Critical patent/CN103208100B/en
Publication of CN103208100A publication Critical patent/CN103208100A/en
Application granted granted Critical
Publication of CN103208100B publication Critical patent/CN103208100B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Abstract

The invention discloses a blurred kernel inversion method for blurred retouching images based on blurred deformation Riemann measure and belongs to the technical field of digital image trusted authentication. Isometries before and after image blurring of logarithm Fourier domains are used, and blurred kernels are recovered from the blurred irrelevant amount before and after the image blurring through Riemann geodesic distance. According to the blurred kernel inversion method, Gaussian blur kernels can be recovered effectively and accurately from the blurred retouching images, the recovered Gaussian blur kernels can be used for manufacturing blurred images from source images and identifying whether the images are subjected to blurred retouching, and the method can be further used for recovering 'clean' images from the blurred images.

Description

The image blurring core inversion method of fuzzy retouching of estimating based on fuzzy indeformable Riemann
Technical field
The invention discloses the image blurring core inversion method of a kind of fuzzy retouching of estimating based on fuzzy indeformable Riemann, belong to digital picture authentic authentication technical field.
Background technology
Along with the fast development of Image Acquisition and picture editting's technology, digital picture has incorporated modern people's life, utilizes image editing software can be easily existing image such as to be retouched, synthesize at the editing operation, produces pleasing picture.These are edited exquisite image and are used for internet, Digital Media document, social media etc., are expressing individual character, are beautifying mediaspace and enrich the aspects such as people's spiritual exchange field and bringing into play irreplaceable effect.Meanwhile, be used to the aspects such as advertisement, media, internet through editor's image, gain public trust by cheating, reduced the public credibility of people to Digital Media, causing trust crisis.Therefore, it is very urgent that research digital picture is forged detection, studies especially quantitatively image forge detection and have more major and immediate significance.
Image blurring noise and editor's generation from obtaining equipment, or the display quality of imaged image, or in order to retouch editing trace, be the research emphasis of image processing and computer graphics always.Image blurring retouching is carried out convolution algorithm to eliminate noise, weaken details, to manufacture special efficacy etc. by image and fuzzy operator, is important inpainting operation.The types such as image blurring retouching comprises Gaussian Blur, average is fuzzy, boxlike is fuzzy, motion blur.The fuzzy part existing in image is in the passive generation of acquisition process, and also having part is initiatively to retouch generation in order to reach content consistency.The key of image deblurring is to recover fuzzy core from blurred picture.Classic method utilization has or prior imformation is carried out image deblurring and deconvoluted by methods such as energy minimization, partial differential equation, Markov fields, and the Given information entropy providing is provided deblurring effect.Image deblurring problem is not still outstanding issue.
Summary of the invention
Technical matters to be solved by this invention is to overcome the deficiencies in the prior art, provide a kind of fuzzy retouching of estimating based on fuzzy indeformable Riemann image blurring core inversion method, utilize the image blurring forward and backward isometry that meets of logarithm Fourier domain, adopt the image blurring forward and backward fuzzy invariant of Riemann's geodesic distance tolerance, can inverting obtain Gaussian Blur kernel function, thereby provide solid foundation for the qualification of image deblurring and image forge.
The present invention specifically solves the problems of the technologies described above by the following technical solutions:
The image blurring core inversion method of fuzzy retouching of estimating based on fuzzy indeformable Riemann, described fuzzy retouching image is obtained through Fuzzy Processing by source images, and the method comprises the following steps:
Steps A, by source images I and fuzzy retouching image I blurbe converted into respectively logarithm Fourier, the source images after conversion and fuzzy retouching image be designated as respectively into with
Step B, calculate respectively and the Riemann's geodesic distance between vector of unit length v ; The expression formula of described vector of unit length v is as follows:
v = - 2 π 3 ξ 2 ,
In formula, ξ represents logarithm Fourier transform frequency;
Step C, calculate the Gaussian Blur core δ of described Fuzzy Processing according to following formula:
δ = 2 3 | ( Q [ I ~ ] - Q [ I ~ blur ] ) | .
Described by source images I and fuzzy retouching image I blurbe converted into respectively logarithm Fourier, specifically in accordance with the following methods:
First to source images I and fuzzy retouching image I blurcarry out respectively Fourier transform, obtain respectively with ; For the image after Fourier transform with , first carry out following processing: if its real part is zero, its real part is replaced to one and be greater than zero infinitesimal real number; Then to after treatment with , ask for respectively the natural logarithm of its mould, obtain being converted into source images and the fuzzy retouching image of logarithm Fourier, be designated as respectively with .
The fuzzy core inversion method of fuzzy retouching image that the present invention proposes, utilizes the image blurring forward and backward isometry of logarithm Fourier domain, recovers fuzzy core by Riemann's geodesic distance from image blurring forward and backward fuzzy irrelevant amount.The inventive method can effectively and exactly recover Gaussian Blur core from fuzzy retouching image, and the Gaussian Blur core recovering can either be used for manufacturing blurred picture from source images, and whether qualification image is by fuzzy retouching; Can be used in again from blurred picture and recover " totally " image.
Brief description of the drawings
Fig. 1 is logarithm fourier space function track and normal trajectories schematic diagram thereof;
Fig. 2 is the inventive method schematic flow sheet;
Fig. 3 is that the Gaussian Blur of actual use is retouched the difference between core and the Gaussian Blur core of the inventive method inverting.
Embodiment
Below in conjunction with accompanying drawing, technical scheme of the present invention is elaborated:
Image is as a kind of 2D signal, and it can represent at multiple transform domains, for example frequency domain (Fourier transform), Laplace domain etc.Gaussian Blur essence is image filtering, uses gaussian kernel function (normal distribution) to calculate fuzzy matrix, and uses fuzzy matrix and source images to carry out convolution algorithm, carries out fuzzy.Fuzzy operation can be expressed as the convolution of source images I (x, y) and Gaussian Blur kernel function G (x, y, δ) in time domain, be shown below.
I blur ( x , y ) = I ( x , y ) * G ( x , y , δ )
In formula δ is at the standard deviation of gaussian kernel normal distribution (referred to as Gaussian Blur core), I blur(x, y) is fuzzy retouching image, and * is convolution algorithm.
If f is R → R function, R is real number field, K δfor convolution kernel function k δmeet normalization character, i.e. ∫ K δ(x) d δ=1.If * is convolution operator, K δto the convolution operation of f suc as formula shown in (1).
( f , K δ ) ( x ) = ∫ - ∞ ∞ f ( y ) K δ ( x - y ) dy - - - ( 1 )
From formula (1), thereby convolution kernel K δform half group space R of an isomorphism +.R +in all convolution kernels produce a convolution track [f]={ (f, K after acting on f δ) | δ ∈ R +.
If for the Fourier transform of f, K δfourier transform be .
K ^ δ ( ξ ) = 1 δ ∫ - ∞ ∞ e - π x 2 / δ e - ixξ dx = 1 δ K δ - 1 ( ξ )
Spatial domain convolution operation f*K δbe converted at frequency domain with product .Convolution shows as the property taken advantage of operation at frequency domain, convolution can be converted to additivity operation by getting further logarithm operation.
( f ^ , K ^ δ ) ( ξ ) = f ^ · K ^ δ - - - ( 2 )
If f 1, f 2for any two signals, with be respectively f 1and f 2with core convolution, in logarithm fourier space, with between Riemann's geodesic distance be calculated as follows shown in.
| | f ~ 1 / - f ~ 2 / | | = | | ( f ~ 1 - δ 0 π ξ 2 ) - ( f ~ 2 - δ 0 π ξ 2 ) | | = | | f ~ 1 - f ~ 2 | | - - - ( 3 )
for the logarithm Fourier transform of signal f.After above formula explanation logarithm fourier space convolution, the forward and backward Riemann's geodesic distance of any two signal convolution equates, not affected by convolution; Show that logarithm fourier space Riemann geodesic distance can be used as the tolerance that measures convolution signal.
In logarithm fourier space, f and f briemann's geodesic distance suc as formula shown in (4).
| | f ~ - f ~ b | | = | | f ~ - ( f ~ - πδ ξ 2 ) | | = | | πδ ξ 2 | | - - - ( 4 )
F refers to original signal, f brefer to Gaussian Blur signal.Norm in formula adopts index Riemann metric to calculate.
Along side-play amount π ξ 2the vector of unit length of trajectory direction is:
v = - π ξ 2 ⟨ π ξ 2 , π ξ 2 ⟩ = - ξ 2 ∫ ξ 4 e - π ξ 2 dξ = - 2 π 3 ξ 2 ;
True origin with the vector of upper any point .As shown in Figure 1, in logarithm fourier space, track and side-play amount π ξ 2track to meet isometry constraint, must make function vertically ; Whole on trace, Q ( f ~ ) = ⟨ ⟨ f ~ , v ⟩ ⟩ .
In Fig. 1, trace-π ξ 2it is δ=1 o'clock convolution kernel represent trace corresponding to convolution kernel that δ is different.Correspondingly, each K δcorresponding difference, there is unique definite solution.In the situation that δ is definite, edge motion, there will be be greater than, equal and be less than 03 kinds of situations. be greater than the minus cut off value of zero-sum and be required δ.
Edge motion f and K δdependence can be by unified formula express, wherein due to function integral form function, so Q ( f ~ - δ ~ π ξ 2 ) = Q ( f ~ ) + δ ~ Q ( π ξ 2 ) = C . Q (π ξ 2) in the integrated value in R territory be , therefore formula (5) is set up.
Q ( f ~ ) + δ ~ 3 2 = C - - - ( 5 )
Obtain from formula (5) .The C of formula (5) can use the convolution f of f bquantization function replace.
According to above analysis, can obtain the image blurring core inversion method of the fuzzy retouching of estimating based on fuzzy indeformable Riemann of the present invention, specific as follows:
Steps A, in accordance with the following methods by source images I and fuzzy retouching image I blurbe converted into respectively logarithm Fourier:
First to source images I and fuzzy retouching image I blurcarry out respectively Fourier transform, obtain respectively with ; For the image after Fourier transform with , first carry out following processing: if its real part is zero, its real part is replaced to one and be greater than zero infinitesimal real number; Then to after treatment with , ask for respectively the natural logarithm of its mould (being square root sum square of real part and imaginary part), obtain being converted into source images and the fuzzy retouching image of logarithm Fourier, be designated as respectively with ;
Step B, calculate respectively and the Riemann's geodesic distance between vector of unit length v ; The expression formula of described vector of unit length v is as follows:
v = - 2 π 3 ξ 2 ,
In formula, ξ represents logarithm Fourier transform frequency;
Step C, calculate the Gaussian Blur core δ of described Fuzzy Processing according to following formula:
δ = 2 3 | ( Q [ I ~ ] - Q [ I ~ blur ] ) | .
In order to verify effect of the present invention, utilize the fuzzy retouching core of the inventive method inversion chart picture and then fuzzy retouching forged to image and detect.Under MATLAB2010 environment, realize fuzzy retouching inversion algorithm of the present invention.Experimental Hardware platform is: four core I7 processors, 8G internal memory.Image source data are from CASIA image set, and picture size is 384 × 256.
The fuzzy retouching image using in experiment is edited generation by convolution function function and the image editing software PHOTOSHOP of MATLAB.
Specific experiment method is as follows:
Use the different source images of 4 width, in Gaussian Blur core δ=0.4, δ=0.6, δ=0.8 and δ=1.0 situation, source images is carried out to Gaussian Blur retouching respectively, and adopt the inventive method to carry out inverting to Gaussian Blur core.Fig. 3 has shown the difference between Gaussian Blur retouching core and the Gaussian Blur core of the inventive method inverting of actual use.As can be seen from the figure, along with the increase of Gaussian Blur core, image blurring degree aggravation, the error that the Gaussian Blur core that algorithm recovers and actual Gauss used retouch core remains on 0.1 left and right, illustrates that the inventive method can recover more exactly Gaussian Blur core from fuzzy retouching image.

Claims (1)

1. the image blurring core inversion method of fuzzy retouching of estimating based on fuzzy indeformable Riemann, described fuzzy retouching image
Obtained through Fuzzy Processing by source images, it is characterized in that, the method comprises the following steps:
Steps A, by source images with fuzzy retouching image be converted into respectively logarithm Fourier, the source images after conversion and fuzzy retouching image are designated as respectively with ; Described by source images with fuzzy retouching image be converted into respectively logarithm Fourier, specifically in accordance with the following methods:
First to source images with fuzzy retouching image carry out respectively Fourier transform, obtain respectively with ; For the image after Fourier transform with , first carry out following processing: if its real part is zero, its real part is replaced to one and be greater than zero infinitesimal real number; Then to after treatment with , ask for respectively the natural logarithm of its mould, obtain being converted into source images and the fuzzy retouching image of logarithm Fourier, be designated as respectively with ;
Step B, calculate respectively , with vector of unit length between Riemann's geodesic distance , ; Described vector of unit length expression formula as follows:
In formula, represent logarithm Fourier transform frequency;
Step C, calculate the Gaussian Blur core of described Fuzzy Processing according to following formula :
CN201310089605.8A 2013-03-19 2013-03-19 Blurred kernel inversion method for blurred retouching images based on blurred deformation Riemann measure Active CN103208100B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201310089605.8A CN103208100B (en) 2013-03-19 2013-03-19 Blurred kernel inversion method for blurred retouching images based on blurred deformation Riemann measure

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201310089605.8A CN103208100B (en) 2013-03-19 2013-03-19 Blurred kernel inversion method for blurred retouching images based on blurred deformation Riemann measure

Publications (2)

Publication Number Publication Date
CN103208100A CN103208100A (en) 2013-07-17
CN103208100B true CN103208100B (en) 2014-08-20

Family

ID=48755317

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201310089605.8A Active CN103208100B (en) 2013-03-19 2013-03-19 Blurred kernel inversion method for blurred retouching images based on blurred deformation Riemann measure

Country Status (1)

Country Link
CN (1) CN103208100B (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7397964B2 (en) * 2004-06-24 2008-07-08 Apple Inc. Gaussian blur approximation suitable for GPU
US7466870B2 (en) * 2003-01-16 2008-12-16 Acoustic Technology Llc Apparatus and method for creating effects in video
CN102013101A (en) * 2010-11-27 2011-04-13 上海大学 Blind detection method of permuted and tampered images subjected to fuzzy postprocessing
CN102750679A (en) * 2012-06-28 2012-10-24 西安电子科技大学 Blind deblurring method for image quality evaluation
CN102903078A (en) * 2012-07-13 2013-01-30 南京大学 motion-blurred image parameter estimation method based on multi-resolution Fourier analysis theory

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7466870B2 (en) * 2003-01-16 2008-12-16 Acoustic Technology Llc Apparatus and method for creating effects in video
US7397964B2 (en) * 2004-06-24 2008-07-08 Apple Inc. Gaussian blur approximation suitable for GPU
CN102013101A (en) * 2010-11-27 2011-04-13 上海大学 Blind detection method of permuted and tampered images subjected to fuzzy postprocessing
CN102750679A (en) * 2012-06-28 2012-10-24 西安电子科技大学 Blind deblurring method for image quality evaluation
CN102903078A (en) * 2012-07-13 2013-01-30 南京大学 motion-blurred image parameter estimation method based on multi-resolution Fourier analysis theory

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
基于测地距离的图像滤波方法;王爱齐;《大连交通大学学报》;20120430;第33卷(第2期);第1节 *
王爱齐.基于测地距离的图像滤波方法.《大连交通大学学报》.2012,第33卷(第2期),
部分模糊核已知的混合模糊图像复原算法;陈曦等;《计算机辅助设计与图形学学报》;20120229;第22卷(第2期);全文 *
陈曦等.部分模糊核已知的混合模糊图像复原算法.《计算机辅助设计与图形学学报》.2012,第22卷(第2期),

Also Published As

Publication number Publication date
CN103208100A (en) 2013-07-17

Similar Documents

Publication Publication Date Title
Liu et al. Spatial-phase shallow learning: rethinking face forgery detection in frequency domain
CN112884671B (en) Fuzzy image restoration method based on unsupervised generation countermeasure network
CN103208104B (en) A kind of image de-noising method based on nonlocal theory
CN103020916A (en) Image denoising method combining two-dimensional Hilbert transform and BEMD
Lahmiri Image denoising in bidimensional empirical mode decomposition domain: the role of Student's probability distribution function
Xiong et al. Fault diagnosis of a rolling bearing based on the wavelet packet transform and a deep residual network with lightweight multi-branch structure
CN104008537A (en) Novel noise image fusion method based on CS-CT-CHMM
Patel et al. Separated component-based restoration of speckled SAR images
Quan et al. Homotopic gradients of generative density priors for MR image reconstruction
CN104252704A (en) Total generalized variation-based infrared image multi-sensor super-resolution reconstruction method
Zuo et al. Frequency-dependent depth map enhancement via iterative depth-guided affine transformation and intensity-guided refinement
Chen et al. Removing Gaussian noise for colour images by quaternion representation and optimisation of weights in non‐local means filter
CN107742278B (en) Binding of L0Motion blurred image blind restoration method based on norm and spatial scale information
Deeba et al. Lossless digital image watermarking in sparse domain by using K‐singular value decomposition algorithm
Onur Improved image denoising using wavelet edge detection based on Otsu’s thresholding
Guérin et al. Gradient terrain authoring
Luo et al. Image universal steganalysis based on best wavelet packet decomposition
Shen et al. Mutual information-driven triple interaction network for efficient image dehazing
CN103208100B (en) Blurred kernel inversion method for blurred retouching images based on blurred deformation Riemann measure
CN105303538A (en) Gauss noise variance estimation method based on NSCT and PCA
Bai et al. Image denoising via an improved non‐local total variation model
Yao et al. Multiscale residual fusion network for image denoising
CN103345727B (en) A kind of method for reconstructing of binary optical image spectrum
CN102156967B (en) Multiscale-based local image interpolation method
Chen et al. A new wavelet hard threshold to process image with strong Gaussian Noise

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant