CN103796017B - Image discriminating device and method - Google Patents
Image discriminating device and method Download PDFInfo
- Publication number
- CN103796017B CN103796017B CN201410031464.9A CN201410031464A CN103796017B CN 103796017 B CN103796017 B CN 103796017B CN 201410031464 A CN201410031464 A CN 201410031464A CN 103796017 B CN103796017 B CN 103796017B
- Authority
- CN
- China
- Prior art keywords
- image
- coefficient
- coefficient block
- block set
- dct
- 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
Links
Landscapes
- Compression Of Band Width Or Redundancy In Fax (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention discloses an image discriminating device and method. The device comprises a first coefficient block set generating module, a second coefficient block set generating module, a DCT quantized noise set generating module and a discriminating module, wherein the first coefficient block set generating module is used for segmenting an image and removing subblocks with pixel values being 0 or 255, and DCT conversion is conducted to obtain a first coefficient block set; the second coefficient block set generating module is used for rounding DCT coefficient values in the first coefficient block set and removing all the subblocks with pixel values being 0, and then a second coefficient block set is formed; the DCT quantized noise set generating module is used for obtaining difference values between coefficient values at different positions in the first coefficient block set and the second coefficient block set, and then a DCT quantized noise set is formed; the discriminating module is used for calculating variance of coefficient values in the DCT quantized noise set, and the image is discriminated as a JPEG uncompressed image if the variance is lower than a threshold value. Whether a high-compression-quality image which can not be discriminated in the prior art is a JPEG uncompressed image can be effectively discriminated through the image discriminating device and method.
Description
Technical field
The present invention relates to technical field of image processing, particularly relate to a kind of for differentiating image whether jpeg decompression contract drawing picture
Image discriminating device and method of discrimination.
Background technology
JPEG (Joint Photographic Experts Group) is the picture format being widely used in many occasions, because of
It is easily stored and transmits and is extensively applied by a large amount of.Jpeg image compression standard is by International Telecommunication Union (International
Telecommunication Union, ITU), International Organization for Standardization (International Organization for
Standardization, ISO) and International Electrotechnical Commission (International Electrotechnical
Commisson, IEC) proposed first in 1988, and obtain rapidly popularization and application.Its encoding-decoding process is respectively such as Fig. 1 and Fig. 2
Shown in, wherein quantizer is to represent to utilize self-defining quantization step to remove 64 DCT (Discrete Cosine
Transform) conversion coefficient, then round, FDCT and IDCT represents forward direction dct transform and inverse dct transform respectively.
Along with the development of computer technology and a large amount of of digital camera popularize, people can shoot more and touch greatly
The picture of amount, but simultaneously because the appearance of various easily operated image editing software, the picture more be tampered, forged is the biggest
Amount is emerged in large numbers.Investigate for image's authenticity, during evidence obtaining, especially differentiate the needs of image true-false, all make us want
Grasp the ability distinguishing image true-false.Wherein, differentiate that image whether jpeg decompression contract drawing picture has for the detection of distorted image
Important effect.But distorting and the most visually cannot produce obvious difference, prior art also cannot be the most right
Jpeg decompression contract drawing picture differentiates, such as it is shown on figure 3, work as image uncompressed image I by P1、P2Composition, P1For uncompressed figure
Picture, P2For jpeg decompression contract drawing picture, if P2During compression quality QF=100 or 99, prior art is cannot to differentiate image I
's.
Therefore, prior art has yet to be improved and developed.
Summary of the invention
In view of above-mentioned the deficiencies in the prior art, it is an object of the invention to provide a kind of image discriminating device and differentiation side thereof
Method, it is intended to solve present problems.
Technical scheme is as follows:
A kind of image discriminating device, wherein, described device includes with lower part:
First coefficient block set generation module, is used for dividing the image into into continuous nonoverlapping sub-block, searches and removes and contains
The sub-block having pixel value to be 0 or 255, carries out dct transform respectively and obtains the first coefficient block set remaining sub-block；
Second coefficient block set generation module, for carrying out all DCT coefficient values in described first coefficient block set
Round, search and remove the sub-block of DCT coefficient value all 0 in the first coefficient block set after rounding, form the second coefficient block
Set；
DCT quantizing noise set generation module, for by phase in described first coefficient block set and the second coefficient block set
The DCT coefficient value of correspondence position takes difference, forms DCT quantizing noise set；
Determination module, for calculating the variance yields of all coefficient values in described DCT quantizing noise set, by this variance yields with
The threshold value preset compares, if described variance yields is less than threshold value, then process decision chart picture is jpeg decompression contract drawing picture.
Described image discriminating device, wherein, described device also includes:
Image type differentiates and modular converter, carries out the judgement of image type before image discriminating, when image is that RGB is color
During color image, transition diagram picture is YCbCr image.Y-component is judged.When image is gray level image, directly use gray scale
Value.
A kind of image discriminating method utilizing image discriminating device as above to carry out, wherein, described method include with
Lower step:
A, described first coefficient block set generation module divide the image into into continuous nonoverlapping sub-block, search and remove and contain
The sub-block having pixel value to be 0 or 255, carries out dct transform respectively and obtains the first coefficient block set remaining sub-block；
All DCT coefficient values in described first coefficient block set are carried out by B, described second coefficient block set generation module
Round, search and remove the sub-block of DCT coefficient value all 0 in the first coefficient block set after rounding, form the second coefficient block
Set；
C, described DCT quantizing noise set generation module are by phase in described first coefficient block set and the second coefficient block set
The DCT coefficient value of correspondence position takes difference, forms DCT quantizing noise set；
D, described determination module calculate the variance yields of all coefficient values in described DCT quantizing noise set, and by this variance
Value compares with the threshold value preset, if described variance yields is less than threshold value, then process decision chart picture is jpeg decompression contract drawing picture.
Described image discriminating method, wherein, also includes before described step A: described image type differentiates and modulus of conversion
Block judges image type, and when image is RGB color image, transition diagram picture is YCbCr image.
Described image discriminating method, wherein, described step A particularly as follows:
A1, described first coefficient block set generation module divide the image into into the son of the most nonoverlapping 8 × 8 pixel sizes
Block, searches and removes containing the sub-block that pixel value is 0 or 255, remaining sub-block converted by the sequence of original sequencing
Front set；
A2, described first coefficient block set generation module carry out dct transform respectively to sub-block in set before conversion and obtain the
One coefficient block set.
Described image discriminating method, wherein, described step B particularly as follows:
All DCT coefficient values in described first coefficient block set are led to by B1, described second coefficient block set generation module
Cross its fractional part to have rounded up rounding operation, obtain rounding rear coefficient block set；
B2, described second coefficient block set generation module round in rear coefficient block set 2-64 described in searching and removing
The sub-block of coefficient value all 0, residue sub-block forms the second coefficient block set.
Described image discriminating method, wherein, described step C particularly as follows:
C1, described DCT quantizing noise set generation module are by described first coefficient block set and described second coefficient block collection
Conjunction is compared, and retains sub-block corresponding with described second coefficient block collection location in described first coefficient block set and forms the 3rd
Coefficient block set；
C2, described DCT quantizing noise set generation module are by described 3rd coefficient block set and the second coefficient block set
The DCT coefficient value of opposite position takes difference, and by the sequence of original sequencing, all differences are formed DCT quantizing noise set.
Described image discriminating method, wherein, if described variance yields is more than or equal to threshold value in described step D, then described
Determination module process decision chart picture is uncompressed image.
Described image discriminating method, wherein, described threshold value is 0.069.
Described image discriminating method, wherein, described first coefficient block set generation module divides the image into into the most not
The sub-block of 8 × 8 overlapping pixel sizes, when remainder occur after segmentation, and this remainder is less than 8 × 8 pixel sizes, then
Remove this remainder.
Beneficial effect: the present invention provides a kind of J image discriminating device and method of discrimination thereof, by apparatus of the present invention and method
Achieve effectively the differentiating as image dimerous by uncompressed image and jpeg decompression contract drawing, it is possible to picture format
It is made to determine whether to be contracted by jpeg decompression to save as current lattice again for unpressed image (such as tiff format, BMP form etc.)
The image of formula, the image with high compression quality that cannot differentiate especially for prior art, the method for the present invention can effectively be reflected
Whether it is not jpeg decompression contract drawing picture.Additionally the image discriminating method of the present invention is in the differentiation of small size (such as 32 × 32) image
There is in work more sane performance.
Accompanying drawing explanation
Fig. 1 is jpeg image compression process schematic diagram in prior art.
Fig. 2 is jpeg image decompression process schematic diagram in prior art.
Fig. 3 is the composition schematic diagram of image I in prior art.
Fig. 4 is the theory diagram of image discriminating device in the specific embodiment of the invention.
Fig. 5 is image discriminating method flow chart in the specific embodiment of the invention.
Fig. 6 is the concrete grammar flow chart of step S100 in Fig. 5.
Fig. 7 is image segmentation schematic diagram in image discriminating method of the present invention.
Fig. 8 is the schematic diagram that in image discriminating method of the present invention, remainder occurs in image segmentation.
Fig. 9 is to remove sub-block and residue sub-block sequence schematic diagram in image discriminating method of the present invention.
Figure 10 is the DCT coefficient value chronological order schematic diagram of the present invention.
Figure 11 is the concrete grammar flow chart of step S300 in Fig. 5.
Figure 12 is the concrete grammar flow chart of step S300 in Fig. 5.
Figure 13 is that in the specific embodiment of the invention, JPEG compression process produces DCT quantizing noise schematic diagram.
Figure 14 is the method flow diagram that in the specific embodiment of the invention, image discriminating device carries out image discriminating.
Figure 15 is image discriminating device image block result schematic diagram in the embodiment of the present invention 1.
Figure 16 is that in the embodiment of the present invention 1, image discriminating device carries out result schematic diagram after dct transform to sub-block.
Figure 17 is that in the embodiment of the present invention 1, sub-block is rounded by image discriminating device, and removing AC coefficient is 0 sub-block the knot that sorts
Really schematic diagram.
Figure 18 is the discrimination precision result schematic diagram in embodiment 4 to uncompressed image Yu jpeg compressed image.
Detailed description of the invention
The present invention provides a kind of image discriminating device and method of discrimination thereof, for making the purpose of the present invention, technical scheme and effect
Fruit is clearer, clear and definite, and the present invention is described in more detail below.Only should be appreciated that specific embodiment described herein
Only in order to explain the present invention, it is not intended to limit the present invention.
A kind of image discriminating device as shown in Figure 4, wherein, described device includes with lower part:
First coefficient block set generation module 100, is used for dividing the image into into continuous nonoverlapping sub-block, searches and remove
Containing the sub-block that pixel value is 0 or 255, remaining sub-block is carried out respectively dct transform and obtains the first coefficient block set.
Second coefficient block set generation module 200, for entering all DCT coefficient values in described first coefficient block set
Row rounds, and searches and removes the sub-block of DCT coefficient value all 0 in the first coefficient block set after rounding, form the second coefficient
Set of blocks.
DCT quantizing noise set generation module 300, for by described first coefficient block set and the second coefficient block set
The DCT coefficient value of opposite position takes difference, forms DCT quantizing noise set.
Determination module 400, for calculating the variance yields of all coefficient values in described DCT quantizing noise set, by this variance
Value compares with the threshold value preset, if described variance yields is less than threshold value, then process decision chart picture is jpeg decompression contract drawing picture.
In preferred embodiment, described device also includes:
Image type differentiates and modular converter 500, carries out the judgement of image type, when image is before image discriminating
During RGB color image, transition diagram picture is YCbCr image, and uses luminance component Y.When image is gray-scale map, directly initiate
One coefficient block set generation module performs corresponding operating.
It is illustrated in figure 5 the image discriminating method of described image discriminating device, wherein, said method comprising the steps of:
S100, described first coefficient block set generation module divide the image into into continuous nonoverlapping sub-block, search and go
Except containing the sub-block that pixel value is 0 or 255, remaining sub-block being carried out respectively dct transform and obtains the first coefficient block set.
In preferred embodiment, also include before described step S100: described image type differentiates and modular converter process decision chart
As type, when image is gray-scale map, the most directly performing step S100, when image is RGB color image, transition diagram picture is
YCbCr image, uses luminance component Y.
Further, described step S100 as shown in Figure 6, particularly as follows:
S110, described first coefficient block set generation module divide the image into into the most nonoverlapping 8 × 8 pixel sizes
Sub-block, searches and removes containing the sub-block that pixel value is 0 or 255, remaining sub-block become by the sequence of original sequencing
Change front set.
It is preferred that when remainder occur after segmentation, and this remainder is less than 8 × 8 pixel sizes, then remove this and remain
Remaining part is divided.
Concrete processing procedure is as it is shown in fig. 7, carry out image I being divided into continuous nonoverlapping sub-block, and each sub-block is (such as
I1) size is 8 × 8 pixels.As shown in Figure 8, for finally segmenting sub-block I2After, remainder P pixel value is less than 8 × 8 pixels
Value size, the most directly remove this portion P, statistics is split obtains K1Individual sub-block.J(i)(i=1,2 ..., K1) represent i-th
Sub-block, for ease of mark, is described as its sub-block combinationsIt is set before conversion.
Owing to, during image is carried out jpeg decompression contracting, the coefficient less than 0 or more than 255 all can be truncated into 0
Or 255, thus can produce error (or referred to as truncation noise).For avoiding the truncation noise produced to affect the differentiation of the inventive method
Performance, removes the sub-block that pixel value in J is 0 or 255, and remaining sub-block quantity is K2, then K2≤K1.Effective sub-block will be remained
According to original sequencing rearrangement, obtainThe most as shown in Figure 9, it is assumed for example that in sub-block 5
Containing 255 pixels.Containing 0 pixel in sub-block 7, then sub-block 5 and sub-block 7 are deleted, by remaining sub-block according to original order
Rearrangement.
S120, described first coefficient block set generation module carry out dct transform respectively to sub-block in set before conversion and obtain
First coefficient block set.
Concrete, before described conversion, set is J(i)(i=1,2 ..., K2), respectively to J(i)(i=1,2 ..., K2) execute
Row dct transform, obtains the first coefficient block setWherein, i represents block index value, and m represents frequency rope
Draw value,Represent m-th coefficient value in i-th sub-block, wherein, as m=0,Represent DC coefficient (DC) value, when 1≤
M≤63,Represent ac coefficient (AC) value.DCT coefficient value coded sequence as shown in Figure 10, to 8 × 8 coefficients after dct transform
Value block (such as form 10) carries out coefficient value layout, uses "the" shape sequential organization coefficient value, from the beginning of the upper left corner, to the lower right corner
Terminate.Form layout pixel value table as shown in Table 20, numerical value m in table (m=0,1,2 ..., 63), represent layout respectively
Rear m-th pixel value, wherein, No. 0 element i.e. represents direct current (DC) coefficient after dct transform, remaining 63 element representation exchange
(AC) coefficient.Numbering in Figure 10 from 0～63 (i.e. m values), actual corresponding order is from the 1st to the 64th, namely No. 0 element pair
Answer the 1st coefficient, No. 1 corresponding 2nd coefficient of element ... No. 63 corresponding 64th coefficients of element.
S200, described second coefficient block set generation module are by all DCT coefficient values in described first coefficient block set
Rounding, search and remove the sub-block of DCT coefficient value all 0 in the first coefficient block set after rounding, forming second is
Several piece set.
In preferred embodiment, described step S200 as shown in figure 11, particularly as follows:
S210, described second coefficient block set generation module are by all DCT coefficient values in described first coefficient block set
Rounded up rounding operation by its fractional part, obtained rounding rear coefficient block set.I.e. to described(i=1,
2 ..., K2The fractional part round computing of all of DCT coefficient value in).Obtaining new coefficient value block is
S220, described second coefficient block set generation module round 2-64 in rear coefficient block set described in searching and removing
The sub-block of individual coefficient value all 0, residue sub-block forms the second coefficient block set.Inspect above-mentioned the most one by oneBy 2-64
Individual coefficient is all the sub-block of zero and removes (i.e. remove ac coefficient value and be all the sub-block of 0).Remaining sub-block quantity is K3, obtainThe most as shown in Figure 10, by 1-63 element numbered in table 20, (in corresponding sub-block, 2-64 is individual
Coefficient) be all 0 sub-block remove.
S300, described DCT quantizing noise set generation module are by described first coefficient block set and the second coefficient block set
The DCT coefficient value of middle opposite position takes difference, forms DCT quantizing noise set.
In preferred embodiment, described step S300 as shown in figure 12, particularly as follows:
S310, described DCT quantizing noise set generation module are by described first coefficient block set and described second coefficient block
Set is compared, and retains sub-block corresponding with described second coefficient block collection location in described first coefficient block set and forms the
Three coefficient block set.I.e. utilize effective sub-block.To the set obtained in described step S120Process, only retain with described
Step S220 obtainsThat K that middle position is corresponding3Individual sub-block, obtains new coefficient value blockIt is the 3rd coefficient block set.
S320, described DCT quantizing noise set generation module are by described 3rd coefficient block set and the second coefficient block set
The DCT coefficient value of middle opposite position takes difference, and by the sequence of original sequencing, all differences are formed DCT quantizing noise collection
Close.The most as shown in figure 13, uncompressed image 30 is according to the compression method (i.e. according to method described in S100, S200 step) of the present invention
Process, in quantizing process, rightWithTherebetween the coefficient of opposite position takes difference, i.e. can get DCT and quantifies
Noise N, wherein(1≤i≤K3, 0≤m≤63).Thus obtain comprising K3All correspondences in individual sub-block
The difference of coefficient, altogether K3× 64 elements, this K3Individual sub-block, by original order sequence, obtainsI.e.
For DCT quantizing noise set.
S400, described determination module calculate the variance yields of all coefficient values in described DCT quantizing noise set, and by the party
Difference compares with the threshold value preset, if described variance yields is less than threshold value, then process decision chart picture is jpeg decompression contract drawing picture.
And described variance yields is more than or equal to threshold value, the most described determination module process decision chart picture is uncompressed image.
In preferred embodiment, described threshold value is 0.069.
Concrete, to sub-blocks all in above-mentioned N ' coefficient value, its quantizing noise of the most all of DCT coefficient takes variance.
First all of DCT quantizing noise is taken average, obtainThen variance yields can be obtainedSet threshold value TA, when the variance drawnMore than or equal to this threshold
Value TA, it is determined that given test image is uncompressed image, otherwise, it is jpeg decompression contract drawing picture.I.e.
Carry out the specific embodiment of image discriminating as shown in figure 14 for image discriminating device of the present invention, specifically comprise the following steps that
H1, given image I, be divided into the most nonoverlapping 8 × 8 sub-block of pixels.
Containing the sub-block that pixel value is 0 or 255 in H2, removal I.
H3, piecemeal implement dct transform, and after removing conversion, AC coefficient is all the sub-block of zero.
H4, corresponding coefficient value is taken difference, obtain the set N of DCT coefficient quantizing noise
H5, coefficient value all of in N is taken variance.
H6、IfMore than or equal to TA, it is determined that image I is uncompressed image, otherwise, I is JPEG
Decompressing image.
Below by embodiment, image discriminating device and the method for discrimination thereof of the present invention are further described.
Embodiment 1
Coloured image at UCID one uncompressed tiff format of image library random choose.
The image selected is intercepted into 32 from central authorities by the first coefficient block set generation module of described image discriminating device
The image block of × 32.Carry out 8 × 8 piecemeals, be divided into 16 sub-blocks, as shown in figure 15.Find the 8th, 12 sub-blocks and contain pixel
Value 255, after the 8th, 12 sub-blocks being removed, remaining 14 sub-blocks are by original sequencing sequence, then carry out dct transform,
To DCT coefficient value, as shown in figure 16.
Described second coefficient block set generation module carries out rounding operation to sub-block as shown in figure 16, removes acquired results
In all AC coefficients be the sub-block of zero, and resequence, obtain result as shown in figure 17.
Described DCT quantizing noise set generation module to taking difference between the effective sub-block shown in Figure 17, and by cross institute
Stating determination module and calculate variance, obtaining variance yields isBy this value and described threshold value TA(0.069) compare
Relatively, obtain this variance yields result more than threshold value, therefore draw the result of determination that image is uncompressed image.Result of determination is just
Really.
Embodiment 2
Downloaded the coloured image of a jpeg format by the Internet at random, again carry out according to step described in embodiment 1.
First pass through described image type and differentiate that with modular converter, this jpeg image being preserved form is tiff image and incites somebody to action
Tiff image turns gray value, and described first coefficient block set generation module is to not finding after this image block containing 0 or 255 pictures
The sub-block of element value, then carry out dct transform to all sub-blocks,
Sub-block after conversion is processed by described second coefficient block set generation module, and ac coefficient is not all zero
Sub-block 1,6,7,12,13,14 retains and resequences.
Described DCT quantizing noise set generation module takes difference, by institute to the DCT coefficient value between the sub-block after sequence
Stating discrimination module and calculate variance, obtaining variance yields isAnd compare with described threshold value, because less than threshold
Value TA, then process decision chart picture is jpeg decompression contract drawing picture.Result of determination is correct.
Embodiment 3
Go back to uncompressed form again with the uncompressed JPEG of going to the image discriminating device and method of the present invention is verified.
Two tiff format image TIF1 and TIF2 are selected at random, respectively with compressibility factor Q=99 and Q in UCID image library
=100 are compressed into jpeg image JPG1 and JPG2；
JPG1 Yu JPG2 solution respectively is pushed back tiff format image, preserves and test according to the step described in embodiment 1
Calculate.Obtaining final result is:
The variance yields of TIF1Less than threshold value TA, therefore image is jpeg decompression contract drawing picture.Judge knot
Fruit is correct.
The variance yields of TIF2Less than threshold value TA, therefore image is jpeg decompression contract drawing picture.Judge knot
Fruit is correct.
Embodiment 4
The image discriminating device and method utilizing 2500 uncompressed images to carry out the present invention is tested.It is translated into ash
Degree figure, and intercept middle body to generate smaller size of image, as pixel be 256 × 256,128 × 128,64 × 64,32 ×
The image of 32.With compressibility factor QF=100,99,98,85,75 and 50, it is compressed the most respectively, then utilizes the present invention
Image discriminating device it is differentiated.Utilize method (W.Luo, J.Huang, and that Weiqi Luo et al. proposes simultaneously
G.Qiu,“JPEG error analysis and its applications to Digital Image Forensics”
IEEE Trans.Inf.Forensics Security, vol.5, no.3, pp.480-491,2010) as a comparison, the two it
Between ratio of precision to as shown in 18, it can be seen that when QF=100 and 99 when, i.e. in the situation that quantization step is little
Under (i.e. step-length is 1 or 2), utilize the differentiation result ratio of precision that the image discriminating device of the present invention and method of discrimination thereof obtain
What Luo et al. proposed has huge lifting.
The present invention provides a kind of image discriminating device and method of discrimination thereof, by apparatus of the present invention and method achieve to by
Uncompressed image and jpeg decompression contract drawing, as effective discriminating of image dimerous, cannot differentiate especially for prior art
The image with high compression quality, the method for the present invention can effectively differentiate whether it is jpeg decompression contract drawing picture.Additionally this
Bright image discriminating method has more sane performance in the differentiation work of small size (such as 32 × 32) image.
It should be appreciated that the application of the present invention is not limited to above-mentioned citing, for those of ordinary skills, can
To be improved according to the above description or to convert, all these modifications and variations all should belong to the guarantor of claims of the present invention
Protect scope.
Claims (10)
1. an image discriminating device, it is characterised in that described device includes with lower part:
First coefficient block set generation module, is used for dividing the image into into continuous nonoverlapping sub-block, searches and remove containing picture
Element value is the sub-block of 0 or 255, remaining sub-block carries out dct transform respectively and obtains the first coefficient block set；
Second coefficient block set generation module, for all DCT coefficient values in described first coefficient block set are rounded,
Search and remove the sub-block of DCT coefficient value all 0 in the first coefficient block set after rounding, form the second coefficient block set；
DCT quantizing noise set generation module, for by corresponding in described first coefficient block set and the second coefficient block set
The DCT coefficient value of position takes difference, forms DCT quantizing noise set；
Determination module, for calculating the variance yields of all coefficient values in described DCT quantizing noise set, by this variance yields with default
Threshold value compare, if described variance yields is less than threshold value, then process decision chart picture is jpeg decompression contract drawing picture.
Image discriminating device the most according to claim 1, it is characterised in that described device also includes:
Image type differentiates and modular converter, carries out the judgement of image type, when image is RGB color figure before image discriminating
During picture, transition diagram picture is YCbCr image.
3. one kind utilizes the image discriminating method that image discriminating device as claimed in claim 1 is carried out, it is characterised in that described
Method comprises the following steps:
A, described first coefficient block set generation module divide the image into into continuous nonoverlapping sub-block, search and remove containing picture
Element value is the sub-block of 0 or 255, remaining sub-block carries out dct transform respectively and obtains the first coefficient block set；
All DCT coefficient values in described first coefficient block set are taken by B, described second coefficient block set generation module
Whole, search and remove the sub-block of DCT coefficient value all 0 in the first coefficient block set after rounding, form the second coefficient block collection
Close；
C, described DCT quantizing noise set generation module are by corresponding in described first coefficient block set and the second coefficient block set
The DCT coefficient value of position takes difference, forms DCT quantizing noise set；
D, described determination module calculate the variance yields of all coefficient values in described DCT quantizing noise set, and by this variance yields with
The threshold value preset compares, if described variance yields is less than threshold value, then process decision chart picture is jpeg decompression contract drawing picture.
Image discriminating method the most according to claim 3, it is characterised in that also include before described step A: described image
Type identification judges image type with modular converter, and when image is RGB color image, transition diagram picture is YCbCr image.
Image discriminating method the most according to claim 3, it is characterised in that described step A particularly as follows:
A1, described first coefficient block set generation module divide the image into into the sub-block of the most nonoverlapping 8 × 8 pixel sizes,
Search and remove containing the sub-block that pixel value is 0 or 255, obtain converting front collection by the sequence of original sequencing by remaining sub-block
Close；
A2, described first coefficient block set generation module carry out dct transform respectively obtain first to sub-blocks in set before conversion
Several piece set.
Image discriminating method the most according to claim 3, it is characterised in that described step B particularly as follows:
All DCT coefficient values in described first coefficient block set are passed through it by B1, described second coefficient block set generation module
Fractional part has rounded up rounding operation, obtains rounding rear coefficient block set；
B2, described second coefficient block set generation module round the 2-64 coefficient in rear coefficient block set described in searching and removing
Being worth the sub-block of all 0, residue sub-block forms the second coefficient block set.
Image discriminating method the most according to claim 3, it is characterised in that described step C particularly as follows:
Described first coefficient block set is entered by C1, described DCT quantizing noise set generation module with described second coefficient block set
Row comparison, retains sub-block corresponding with described second coefficient block collection location in described first coefficient block set and forms the 3rd coefficient
Set of blocks；
C2, described DCT quantizing noise set generation module are relative with in the second coefficient block set by described 3rd coefficient block set
The DCT coefficient value answering position takes difference, and by the sequence of original sequencing, all differences are formed DCT quantizing noise set.
Image discriminating method the most according to claim 3, it is characterised in that if described variance yields is more than in described step D
Or equal to threshold value, the most described determination module process decision chart picture is uncompressed image.
Image discriminating method the most according to claim 3, it is characterised in that described threshold value is 0.069.
Image discriminating method the most according to claim 5, it is characterised in that described first coefficient block set generation module
Divide the image into into the sub-block of the most nonoverlapping 8 × 8 pixel sizes, when after segmentation, remainder, and this remainder occurring
Less than 8 × 8 pixel sizes, then remove this remainder.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410031464.9A CN103796017B (en) | 2014-01-23 | 2014-01-23 | Image discriminating device and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410031464.9A CN103796017B (en) | 2014-01-23 | 2014-01-23 | Image discriminating device and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103796017A CN103796017A (en) | 2014-05-14 |
| CN103796017B true CN103796017B (en) | 2017-01-11 |
Family
ID=50671239
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410031464.9A Active CN103796017B (en) | 2014-01-23 | 2014-01-23 | Image discriminating device and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103796017B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107680386A (en) * | 2017-11-07 | 2018-02-09 | 潘柏霖 | A kind of intelligent traffic monitoring system |
| CN108230411B (en) * | 2017-12-29 | 2021-03-09 | 成都工业学院 | Method and device for detecting tampered image |
| CN115170779B (en) * | 2022-07-18 | 2023-03-28 | 生态环境部环境规划院 | Remote sensing image rendering and network publishing method |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US4849930A (en) * | 1987-02-25 | 1989-07-18 | Westinghouse Electric Corp. | Method of compactly storing digital data |
| CN103067713B (en) * | 2013-01-06 | 2017-04-12 | 中国科学院深圳先进技术研究院 | Method and system of bitmap joint photographic experts group (JPEG) compression detection |
-
2014
- 2014-01-23 CN CN201410031464.9A patent/CN103796017B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN103796017A (en) | 2014-05-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108052980B (en) | Image-based air quality grade detection method | |
| CN110992275B (en) | Refined single image rain removing method based on generation of countermeasure network | |
| CN111445454B (en) | Image authenticity identification method and application thereof in license identification | |
| CN103345758B (en) | Jpeg image region duplication based on DCT statistical nature distorts blind checking method | |
| TWI426774B (en) | A method for classifying an uncompressed image respective to jpeg compression history, an apparatus for classifying an image respective to whether the image has undergone jpeg compression and an image classification method | |
| Yoo et al. | Image restoration by estimating frequency distribution of local patches | |
| JP4486780B2 (en) | Nonlinear quantization and similarity matching method for image information retrieval | |
| CN107657619B (en) | A kind of low-light (level) Forest fire image dividing method | |
| CN103796017B (en) | Image discriminating device and method | |
| CN109118442B (en) | Image enhancement method based on Sobel operator filtering | |
| CN103606136B (en) | Based on the video super resolution of key frame and non local constraint | |
| Lo et al. | Exploring semantic segmentation on the dct representation | |
| CN105740872A (en) | Image feature extraction method and apparatus | |
| CN102547477A (en) | Video fingerprint method based on contourlet transformation model | |
| Fu et al. | Screen content image quality assessment using Euclidean distance | |
| CN110223273B (en) | Image restoration evidence obtaining method combining discrete cosine transform and neural network | |
| CN112153388A (en) | Image compression method, device and related equipment | |
| Ahmed et al. | Contrast-distorted image quality assessment based on curvelet domain features | |
| Wu et al. | Review of imaging device identification based on machine learning | |
| CN108764277A (en) | A kind of non-reference picture Fuzzy Quality evaluation algorithms | |
| Kumari et al. | Image fusion techniques based on pyramid decomposition | |
| CN104796726A (en) | Method for detecting image compression traces | |
| KR100375831B1 (en) | Fast texture - based image database Browsing and Sorting Method | |
| CN102202217A (en) | Joint photographic experts group (JPEG) coding method and equipment | |
| Leykin et al. | Differences of edge properties in photographs and paintings |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20190624 Address after: 518000 No. 144, 4-7 Skirt Towers, Yueweigang Mingyuan (North District), Shekou Street, Nanshan District, Shenzhen City, Guangdong Province Patentee after: Shenzhen Guoxinan Information Technology Co., Ltd. Address before: 518060 3688 Nanhai Road, Nanshan District, Shenzhen, Guangdong Patentee before: Shenzhen University |
|
| TR01 | Transfer of patent right |