CN106485738A - A kind of method and apparatus of the JPEG compression quantization step for estimating bitmap - Google Patents
A kind of method and apparatus of the JPEG compression quantization step for estimating bitmap Download PDFInfo
- Publication number
- CN106485738A CN106485738A CN201510531953.5A CN201510531953A CN106485738A CN 106485738 A CN106485738 A CN 106485738A CN 201510531953 A CN201510531953 A CN 201510531953A CN 106485738 A CN106485738 A CN 106485738A
- Authority
- CN
- China
- Prior art keywords
- bitmap
- estimated
- vector
- arbitrary frequency
- quantization step
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Landscapes
- Compression Of Band Width Or Redundancy In Fax (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The present invention provides a kind of method and apparatus of the JPEG compression quantization step for estimating bitmap, to improve the accuracy rate whether detection bitmap comes via jpeg image conversion.Methods described includes:By to the histogrammic statistics of bitmap DCT coefficient to be estimated, extracting arbitrary frequency f of bitmap to be estimatedi,jCorresponding essential characteristic Ci,j;By arbitrary frequency f that count and combine collimation technique, extract to be estimated bitmap histogrammic to DCT coefficient in bitmap to be estimatedi,jCorresponding supplemental characteristic C 'i,j;Series connection essential characteristic Ci,jWith supplemental characteristic C 'i,j, obtain assemblage characteristic [Ci,j, C 'i,j];All assemblage characteristic [C by bitmap to be estimatedi,j, C 'i,j] input quantization step-size estimation device, obtain the estimate of the JPEG compression quantization step of bitmap to be estimated.The technical scheme that the present invention is provided is higher to the utilization rate of discriminant information when whether detection bitmap is converted by jpeg image, is conducive to obtaining higher accuracy of detection, significantly improves the accuracy rate of the estimation of the JPEG compression quantization step of bitmap.
Description
Technical field
The invention belongs to image processing field, more particularly to a kind of JPEG compression quantization step for estimating bitmap
Method and apparatus.
Background technology
JPEG (Joint Photographic Experts Group, JPEG) form be at present most
Popular internet (Internet) picture format, the occupancy volume in internet are 70.5%, and to JPEG
The conversion of image happens occasionally.After one width jpeg image is changed, may not protected in the jpeg-format
Deposit, but save as bitmap (Bitmap) form.Industry is generally using staying to JPEG compression process
Quantized character and Coefficients Distribution are analyzed, so as to whether judge bitmap through JPEG compression, or even
Estimate the compression parameters adopted during JPEG compression.
A kind of method of existing JPEG compression quantization step for estimating bitmap is discrete remaining using image block
First place numeral (the such as 12 the first number of string conversion (Discrete Cosine Transform, DCT) coefficient
It is 3 that word is 1,321 the first numeral) characteristic carries out JPEG compression detection.This method be based on as follows
The fact, i.e.,:The first digital distribution of the piecemeal DCT coefficient of uncompressed bitmap obeys broad sense Benford
Rule, and the first digital distribution (i.e. statistic histogram) of the bitmap through JPEG compression can then run counter to this
Whether statistical law, can differentiate bitmap to be detected through JPEG compression accordingly.However, the deficiency of the method
Part is to need to collect substantial amounts of DCT coefficient to form stable the first statistics histogram, so as to enter
Row is accurately adjudicated, therefore, when cannot provide enough statistics when bitmap size to be detected is less,
The accuracy rate whether its detection bitmap comes via jpeg image conversion is relatively low.
The method of the existing another kind of JPEG compression quantization step for estimating bitmap is to utilize jpeg compressed image
DCT coefficient meets certain characteristic being distributed to carry out JPEG compression detection.This method be based on another thing
Real, i.e.,:For the bitmap without JPEG compression, its DCT coefficient distribution (i.e. statistic histogram) is substantially
Laplacian distribution is obeyed, and for the bitmap through JPEG compression, its DCT coefficient distribution can be with amount
Change the difference of step-length and present different patterns.In other words, shown not according to DCT coefficient distribution
Same pattern, it can be inferred that the quantization step adopted during JPEG compression.However, the method also has deficiency
Place, for example, in the case that compressive strength is relatively slight, DCT coefficient distribution reflection quantifies the energy of step change
Whether power is weaker, and at this moment the estimation accuracy rate of quantization step is relatively low, so as to detect bitmap via jpeg image
The accuracy rate that conversion comes accordingly reduces.
Content of the invention
It is an object of the invention to provide a kind of method and apparatus of the JPEG compression quantization step for estimating bitmap,
To improve the accuracy rate of the estimation of the JPEG compression quantization step to bitmap, so as to whether improve detection bitmap
Via the accuracy rate that jpeg image conversion comes.
The present invention is achieved in that a kind of method of the JPEG compression quantization step for estimating bitmap, described
Method includes:
By the statistics to discrete cosine transform coefficient histogram in bitmap to be estimated, extract described in wait to estimate
Arbitrary frequency f of meter bitmapi,jCorresponding essential characteristic Ci,j;
Collimation technique is counted and is combined by histogrammic to DCT coefficient in the bitmap to be estimated, extract institute
State arbitrary frequency f of bitmap to be estimatedi,jCorresponding supplemental characteristic C 'i+j, the supplemental characteristic C 'i,jIt is and the base
Eigen Ci,jDimension identical vector;
Connect essential characteristic Ci,jWith the supplemental characteristic C 'i,j, obtain assemblage characteristic [Ci,j, C 'i,j];
All described assemblage characteristic [C by the bitmap to be estimatedi,j, C 'i,j] input quantization step-size estimation device, obtain
Obtain the estimate of the JPEG compression quantization step of the bitmap to be estimated.
Another object of the present invention is to a kind of device of the JPEG compression quantization step for estimating bitmap is provided,
Described device includes:
Essential characteristic extraction module, for by discrete cosine transform coefficient Nogata in bitmap to be estimated
The statistics of figure, extracts arbitrary frequency f of the bitmap to be estimatedi,jCorresponding essential characteristic Ci,j;
Supplemental characteristic extraction module, for by the histogrammic statistics of DCT coefficient in the bitmap to be estimated
And collimation technique is combined, extract arbitrary frequency f of the bitmap to be estimatedi,jCorresponding supplemental characteristic C 'i,j, institute
State supplemental characteristic C 'i,jIt is and essential characteristic Ci,jDimension identical vector;
Assemblage characteristic acquisition module, for essential characteristic C of connectingi,jWith the supplemental characteristic C 'i,j, obtain
Assemblage characteristic [Ci,j, C 'i,j];
Estimate acquisition module, for by all described assemblage characteristic [C of the bitmap to be estimatedi,j, C 'i,j]
Input quantifies step-size estimation device, obtains the estimate of the JPEG compression quantization step of the bitmap to be estimated.
Knowable to the technical scheme that the invention described above is provided, on the one hand, by arbitrary frequency of bitmap to be estimated
fi,jCorresponding essential characteristic Ci,j, i.e., the factor of itself and DCT coefficient is distributed using the DCT coefficient of bitmap
Histogram estimates the JPEG compression quantization step of bitmap, with respect to prior art, detection bitmap whether by
Utilization rate when jpeg image is converted to discriminant information is higher, is conducive to obtaining higher accuracy of detection;
On the other hand, arbitrary frequency f of bitmap to be estimated is extracted in conjunction with calibration certificate technologyi,jCorresponding supplemental characteristic C 'i,j,
That is, its reference picture is obtained by jpeg image collimation technique, and correspondingly carries on these reference pictures
Its DCT coefficient distribution and its factor histogram are taken as supplemental characteristic, significantly improve bitmap further
The accuracy rate of the estimation of JPEG compression quantization step.
Description of the drawings
Fig. 1 is the reality of the method for the JPEG compression quantization step of the estimation bitmap that the embodiment of the present invention one is provided
Existing schematic flow sheet;
Fig. 2 is the offer of the embodiment of the present invention two to the corresponding DCT system of any one frequency in 64 frequencies
The corresponding histogram schematic diagram of any one frequency that the distribution of number absolute values is obtained when being counted;
Fig. 3 is the knot of the device of the JPEG compression quantization step of the estimation bitmap that the embodiment of the present invention three is provided
Structure schematic diagram;
Fig. 4 is the knot of the device of the JPEG compression quantization step of the estimation bitmap that the embodiment of the present invention four is provided
Structure schematic diagram figure;
Fig. 5 is the knot of the device of the JPEG compression quantization step of the estimation bitmap that the embodiment of the present invention five is provided
Structure schematic diagram;
Fig. 6-a is the knot of the device of the JPEG compression quantization step of the estimation bitmap that the embodiment of the present invention six is provided
Structure schematic diagram;
Fig. 6-b is the device of the JPEG compression quantization step of the estimation bitmap that the embodiment of the present invention seven is provided
Structural representation;
Fig. 6-c is the knot of the device of the JPEG compression quantization step of the estimation bitmap that the embodiment of the present invention eight is provided
Structure schematic diagram;
Fig. 7-a is the knot of the device of the JPEG compression quantization step of the estimation bitmap that the embodiment of the present invention nine is provided
Structure schematic diagram;
Fig. 7-b is the device of the JPEG compression quantization step of the estimation bitmap that the embodiment of the present invention ten is provided
Structural representation;
Fig. 7-c is the device of the JPEG compression quantization step of the estimation bitmap that the embodiment of the present invention 11 is provided
Structural representation.
Specific embodiment
In order that the purpose of the present invention, technical scheme and beneficial effect become more apparent, below in conjunction with accompanying drawing
And embodiment, the present invention will be described in further detail.It should be appreciated that described herein be embodied as
Example is not intended to limit the present invention only in order to explain the present invention.
The embodiment of the present invention provides a kind of method of the JPEG compression quantization step for estimating bitmap, methods described
Including:By the statistics to discrete cosine transform coefficient histogram in bitmap to be estimated, extract described in treat
Estimate arbitrary frequency f of bitmapi,jCorresponding essential characteristic Ci,j;By to DCT system in the bitmap to be estimated
The histogrammic statistics of number simultaneously combines collimation technique, extracts arbitrary frequency f of the bitmap to be estimatedi,jCorresponding auxiliary
Help feature C 'i,j, the supplemental characteristic C 'i,jIt is and essential characteristic Ci,jDimension identical vector;Series connection is described
Essential characteristic Ci,jWith the supplemental characteristic C 'i,j, obtain assemblage characteristic [Ci,j, C 'i,j];By the bitmap to be estimated
All described assemblage characteristic [Ci,j, C 'i,j] input quantization step-size estimation device, obtain the bitmap to be estimated
The estimate of JPEG compression quantization step.The embodiment of the present invention also provides the corresponding JPEG pressure for estimating bitmap
The device of contracting quantization step.It is described in detail individually below.
Accompanying drawing 1 is referred to, is the JPEG compression quantization step of the estimation bitmap that the embodiment of the present invention one is provided
Method realize flow process, mainly include the following steps that S101 to step S104:
S101, by the statistics to discrete cosine transform coefficient histogram in bitmap to be estimated, extracts institute
State arbitrary frequency f of bitmap to be estimatedi,jCorresponding essential characteristic Ci,j.
In embodiments of the present invention, bitmap to be estimated refers to that under a cloud come and needed by jpeg image conversion
To be estimated by quantization step being compressed to which, determine whether the position come by jpeg image conversion
Figure.
As one embodiment of the invention, by discrete cosine transform coefficient Nogata in bitmap to be estimated
The statistics of figure, extracts arbitrary frequency f of the bitmap to be estimatedi,jCorresponding essential characteristic Ci,jCan pass through as follows
Step S1011 to S1015 is realized:
S1011, obtains the corresponding DCT system of 64 frequencies of each 8 × 8 resolution ratio sub-block of bitmap to be estimated
Number.
In embodiments of the present invention, the size that can be 8 × 8 pixels according to 8 × 8 resolution ratio is estimated to treat
Meter bitmap carries out piecemeal, and so, each sub-block includes 64 pixels, and 64 pixels correspond to 64 respectively
Individual frequency.Discrete cosine transform (Discrete Cosine Transform, the DCT) coefficient of bitmap to be estimated
Matrix is exactly to be made up of this corresponding DCT coefficient of 64 frequencies.How many DCT are corresponded to as each frequency
Coefficient, this are that the sub-block for being divided into how many 8 × 8 pixel sizes by bitmap to be estimated determines, specifically,
The number of the corresponding DCT coefficient of each frequency is divided into the sub-block of 8 × 8 pixel sizes with bitmap to be estimated
Number identical.For example, if bitmap to be estimated is divided into the sub-block of 32 8 × 8 pixel sizes, per
Individual frequency is just to there is 32 DCT coefficient.
In embodiments of the present invention, the DCT coefficient matrix of bitmap to be estimated can be saved in the form of a file,
Extracting arbitrary frequency f of the bitmap to be estimatedi,jCorresponding essential characteristic Ci,jBefore, first can be from these texts
The corresponding DCT coefficient of 64 frequencies of each 8 × 8 resolution ratio sub-block is obtained in part.
S1012, the distribution of the corresponding DCT coefficient absolute value of 64 each frequencies of frequency of statistics, obtain described
Corresponding 64 histograms of 64 frequencies.
As it was previously stated, when bitmap to be estimated is divided into the sub-block of multiple 8 × 8 pixel sizes, each frequency
Corresponding DCT coefficient has multiple.In embodiments of the present invention, any one frequency in 64 frequencies
When the distribution of the corresponding DCT coefficient absolute value of rate is counted, any one frequency can be obtained corresponding straight
Fang Tu;When the distribution to the corresponding DCT coefficient absolute value of 64 each frequencies of frequency is counted, can
Obtain corresponding 64 histograms of 64 frequencies, i.e. frequency one histogram of correspondence, 64
Frequency correspond to 64 histograms.
Arbitrary frequency f with 64 frequenciesi,jCorresponding DCT coefficient for 0,1,0,15,3,1,
9,3,6,13,0,8,3,10,8,0,13,10,0,9,0,4,13,8,0,
6,1,10,6,4,8,0,15,16,0,12,4,4,9,13,7,10,
0,7,0,15,12,0 } as a example by, these DCT coefficient absolute values for 0,1,0,15,3,1,9,
3,6,13,0,8,3,10,8,0,13,10,0,9,0,4,13,8,0,6,1,10,
6,4,8,0,15,16,0,12,4,4,9,13,7,10,0,7,0,15,12,0 },
When to frequency fi,jThe distribution of these DCT coefficient absolute values corresponding is would know that when being counted:" 0 " occurs
12 times, " 1 " occurs in that 3 times, and " 2 " occur in that 0 time, " 3 " occur in that 3 times, and " 4 " occur
4 times, " 5 " occur in that 0 time, and " 6 " occur in that 3 times, " 7 " occur in that 2 times, and " 8 " occur
4 times, " 9 " occur in that 3 times, and " 10 " occur in that 4 times, " 11 " occur in that " 12 " go out 0 time
Show 2 times, " 13 " occur in that 4 times, " 14 " occur in that 0 time, " 15 " occur in that 3 times, " 16 "
Occur in that 1 time, the description of corresponding histogram be exactly " 0 ", " 1 ", " 2 ", " 3 ", " 4 ",
“5”、“6”、“7”、“8”、“9”、“10”、“11”、“12”、“13”、
The frequency that " 14 ", " 15 " and " 16 " occur, as shown in Figure 2.
It should be noted that in embodiments of the present invention, arbitrary frequency fi,jSubscript i, j represent respectively and wait to estimate
I-th row of the DCT coefficient matrix of meter bitmap, jth row, frequency fi,jExpression DCT coefficient the i-th row of matrix,
Frequency corresponding to the DCT coefficient of jth row.
S1013, intercepts arbitrary frequency f in 64 histogramsi,jL vertical bar in corresponding histogram, obtains institute
State arbitrary frequency fi,jCorresponding vector Gi,j, wherein, vectorial Gi,jFor L dimensional vector.
In embodiments of the present invention, a histogrammic vertical bar (bin) represents by the frequency of occurrence of objects of statistics,
After L vertical bar in histogram is intercepted, the vectorial G for obtainingi,jBe from the L of these frequency constitution elements tie up to
Amount, for example, if intercept 2 example of accompanying drawing histogram in front 10 vertical bars be DCT coefficient absolute value be " 0 ",
" 1 ", " 2 ", " 3 ", " 4 ", " 5 ", " 6 ", " 7 ", " 8 " and " 9 " is corresponding perpendicular
Bar, then arbitrary frequency fi,jCorresponding vector Gi,j={ 12,3,0,3,4,0,3,2,4,3 }.
Finally to the vector Gi,jIt is normalized, will Gi,jDivided by Gi,jThe sum of all elements, be 12/34,
3/34,0,3/34,4/34,0,3/34,2/34,4/34,3/34 }.
Herein it should be noted that intercept 64 histograms in arbitrary frequency fi,jL in corresponding histogram
During individual vertical bar, L value can neither be too small, can not be excessive.This is because, if the acquirement of L value is too small, finally
The accuracy of the estimate of the JPEG compression quantization step of the bitmap to be estimated for obtaining is relatively low, if L value is obtained
Excessive, although the accuracy phase of the estimate of the JPEG compression quantization step of the bitmap to be estimated for finally obtaining
L value is obtained and height is wanted compared with hour, but can increase follow-up Training Support Vector Machines (Support Vector
Machine, SVM) when training difficulty.Therefore, to the value of L should comprehensive each side actual because
Element requires consideration, for example, according to practical experience, in the present invention, the value of L is taken as 100 and compares conjunction
Suitable.
S1014, to vectorial Gi,jEach DCT coefficient absolute value carry out Factorization, Statistics decomposition is obtained
All positive factor value histogram, obtain arbitrary frequency fi,jCorresponding factor histogram feature vector
Fi,j, wherein, factor histogram feature vector Fi,jDimension and the aforementioned vectorial G obtained by step S1013i,j
Dimension identical.
In embodiments of the present invention, to vectorial Gi,jEach DCT coefficient absolute value carry out Factorization,
Can be realized using known technology, not repeat herein.
S1014, connect vector Gi,jWith factor histogram feature vector Fi,j, the result [G of series connectioni,j, Fi,j] be
Arbitrary frequency f of bitmap to be estimatedi,jCorresponding essential characteristic Ci,j.
Obviously, if Gi,jOr Fi,jThe vector of L dimension, then arbitrary frequency f of the bitmap to be estimated for obtaining hereini,j
Corresponding essential characteristic Ci,jThe vector of 2L dimension, for example, when L value is 100, essential characteristic Ci,jIt is
Dimension is 200 vector.
S102, is counted and is combined collimation technique by histogrammic to DCT coefficient in bitmap to be estimated, carry
Take arbitrary frequency f of the bitmap to be estimatedi,jCorresponding supplemental characteristic C 'i,j, wherein, supplemental characteristic C 'i,jBe with
Essential characteristic C obtained by abovementioned steps S101i,jDimension identical vector.
As one embodiment of the invention, by the histogrammic statistics of DCT coefficient in bitmap to be estimated and tying
Collimation technique is closed, extracts arbitrary frequency f of the bitmap to be estimatedi,jCorresponding supplemental characteristic C 'i,jCan pass through
Following steps S1021 to S1026 is realized:
S1021, obtains the reference picture of bitmap to be estimated with collimation technique.
In embodiments of the present invention, the reference picture for obtaining bitmap to be estimated with collimation technique can be walked by following
Rapid S1 to S3 is realized:
S1, the bitmap to be estimated to being input into carry out inverse discrete cosine transformation (Inverse according to 8 × 8 piecemeals
Discrete Cosine Transform, IDCT), picture element matrix is reverted to, wherein, picture element matrix is by many
8 × 8 blocks of pixels composition of non-overlapping copies.If bitmap to be estimated is through JPEG compression, pass through IDCT
The picture element matrix for obtaining generally has blocking effect.
S2, cutting is deleted via 4 row before abovementioned steps S1 gained picture element matrix and front 4 row pixel,
Obtain the picture element matrix after cutting.
If as it was previously stated, bitmap to be estimated is through JPEG compression, the pixel square obtained by IDCT
Battle array generally has blocking effect.Therefore, in order to destroy the blocking effect of original JPEG compression, the embodiment of the present invention
Can cutting or delete via 4 row before abovementioned steps S1 gained picture element matrix and front 4 row pixel.
S3, to carrying out Block DCT (Discrete via the picture element matrix obtained after the cutting of step S2
Cosine Transform, DCT), and the DCT coefficient that conversion is obtained is quantified and entropy code.
Herein, and to the DCT coefficient that obtains of conversion quantified and entropy code after another width bitmap of gained, i.e.,
It is the reference picture of the bitmap to be estimated of the embodiment of the present invention.
S1022, the 64 of each 8 × 8 resolution ratio sub-block of the reference picture that acquisition is obtained via step S1021
The corresponding DCT coefficient of individual frequency.
Step S1022 to realize process similar with S1011 the step of previous embodiment, difference be herein its
Obtained is the corresponding DCT coefficient of 64 frequencies of each 8 × 8 resolution ratio sub-block of reference picture.
S1023, the distribution of the corresponding DCT coefficient absolute value of 64 each frequencies of frequency of statistics, obtain described
Corresponding 64 histograms of 64 frequencies.
Step S1023 to realize process similar with S1012 the step of previous embodiment, difference is herein
64 frequencies are 64 frequencies of reference picture.
S1024, intercepts arbitrary frequency f in 64 histogramsi,jL vertical bar in corresponding histogram, obtains institute
State arbitrary frequency fi,jCorresponding vector G 'i,j, wherein, vectorial G 'i,jFor L dimensional vector.
Step S1024 to realize process similar with S1013 the step of previous embodiment, difference is herein
Histogram is obtained by reference picture.
S1025, connect the vector G 'i,jWith factor histogram feature vector F 'I, j, the result [G ' of the series connectioni,j,
F′i,j] it is arbitrary frequency f of the bitmap to be estimatedi,jCorresponding supplemental characteristic C 'i,j.
S103, essential characteristic C obtained via step S101 of connectingi,jWith the auxiliary obtained via step S102
Feature C 'i,j, obtain assemblage characteristic [Ci,j, C 'i,j].
S104, by all assemblage characteristic [C of bitmap to be estimatedi,j, C 'i,j] input quantization step-size estimation device, obtain
The estimate of the JPEG compression quantization step of the bitmap to be estimated.
In embodiments of the present invention, quantization step estimator is used for entering the JPEG compression quantization step of bitmap
Row estimation, can extract arbitrary frequency f of bitmap to be estimatedi,jCorresponding essential characteristic Ci,jBuild before and complete.
As one embodiment of the invention, arbitrary frequency f of bitmap to be estimated is being extractedi,jCorresponding essential characteristic Ci,jIt
Front structure quantization step estimator can as follows S1041 and step S1042 realizing:
S1041, extracts arbitrary frequency f of the bitmap to be estimated using with aforementionedi,jCorresponding essential characteristic
Ci,jCi,jWith supplemental characteristic C 'i,jIdentical mode, extracts arbitrary frequency f of bitmapi,jCorresponding training assemblage characteristic
Bitmap arbitrary frequency f is extracted hereini,jCorresponding training assemblage characteristicIts mode and previous embodiment
Step S1011 to S1014 extracts arbitrary frequency f of bitmap to be estimatedi,jCorresponding essential characteristic Ci,jSpecial with auxiliary
Levy C 'I, jMode essentially identical, what difference was to extract herein is arbitrary frequency f of training image storehouse Bitmapi,j
Corresponding training assemblage characteristicDo not repeat.
S1042, with arbitrary frequency f obtained through step S1041i,jCorresponding training assemblage characteristicAs
Hold the training characteristics of vector machine (Support Vector Machine, SVM), and using quantization step q as
The assemblage characteristicTraining category, train the SVM to obtain arbitrary frequency fi,jCorresponding quantization step
Long estimator.
Obviously, when the corresponding training assemblage characteristic of each frequency of 64 frequencies is instructed as the training characteristics of SVM
Practice SVM, it is possible to obtain 64 corresponding quantization step estimators.
What step S101 to S103 was obtained is arbitrary frequency f of bitmap to be estimatedi,jCorresponding assemblage characteristic [Ci,j,
C′i,j], the corresponding combination of each frequency of bitmap to be estimated is obtained according to step S101 to S103 similar mode
After feature, you can obtain all assemblage characteristics of bitmap to be estimated.All assemblage characteristics by bitmap to be estimated
After input quantifies step-size estimation device, the output valve of quantization step estimator is exactly the JPEG pressure of bitmap to be estimated
The estimate of contracting quantization step.
Knowable to the method for the JPEG compression quantization step of the estimation bitmap of 1 example of above-mentioned accompanying drawing, on the one hand,
By arbitrary frequency f of bitmap to be estimatedi,jCorresponding essential characteristic Ci,j, i.e., divided using the DCT coefficient of bitmap
The factor histogram of cloth itself and DCT coefficient estimates the JPEG compression quantization step of bitmap, with respect to existing
Technology, is detecting that bitmap is higher to the utilization rate of discriminant information when whether being converted by jpeg image, has
Beneficial to the higher accuracy of detection of acquisition;On the other hand, the arbitrary of bitmap to be estimated is extracted in conjunction with calibration certificate technology
Frequency fi,jCorresponding supplemental characteristic C 'i,j, i.e. its reference picture is obtained by jpeg image collimation technique, and
Its DCT coefficient distribution and its factor histogram are correspondingly extracted on these reference pictures as supplemental characteristic,
The accuracy rate of the estimation of the JPEG compression quantization step of bitmap is significantly improved further.
Accompanying drawing 3 is referred to, is the JPEG compression quantization step of the estimation bitmap that the embodiment of the present invention three is provided
Device can be the executive agent of the method for the JPEG compression quantization step of the estimation bitmap of 1 example of accompanying drawing.
For convenience of description, accompanying drawing 3 illustrate only the part related to the embodiment of the present invention.The estimating of 3 example of accompanying drawing
The device of the JPEG compression quantization step of meter bitmap mainly includes essential characteristic extraction module 301, supplemental characteristic
Extraction module 302, assemblage characteristic acquisition module 303 and estimate acquisition module 304, each functional module are detailed
It is described as follows:
Essential characteristic extraction module 301, for by discrete cosine transform coefficient in bitmap to be estimated
Histogrammic statistics, extracts arbitrary frequency f of the bitmap to be estimatedi,jCorresponding essential characteristic Ci,j;
Supplemental characteristic extraction module 302, for by the histogrammic statistics of DCT coefficient in bitmap to be estimated
And collimation technique is combined, extract arbitrary frequency f of the bitmap to be estimatedi,jCorresponding supplemental characteristic C 'i,j, its
In, supplemental characteristic C 'i,jIt is essential characteristic C that extracts with aforementioned essential characteristic extraction module 301i,jDimension is identical
Vector;
Assemblage characteristic acquisition module 303, for the essential characteristic of the extraction of essential characteristic extraction module 301 of connecting
Ci,jThe supplemental characteristic C ' extracted with supplemental characteristic extraction module 302i,j, obtain assemblage characteristic [Ci,j, C 'i,j];
Estimate acquisition module 304, for by all assemblage characteristic [C of bitmap to be estimatedi,j, C 'i,j] input quantity
Change step-size estimation device, obtain the estimate of the JPEG compression quantization step of bitmap to be estimated.
It should be noted that the device of the JPEG compression quantization step of the estimation bitmap of 3 example of the figures above
Embodiment in, the division of each functional module is merely illustrative of, in practical application can as needed,
The convenient consideration of the realization of the configuration requirement of for example corresponding hardware or software, and above-mentioned functions are distributed by not
With functional module complete, will described estimate bitmap JPEG compression quantization step device internal junction
Structure is divided into different functional modules, to complete all or part of function described above.And, actual
In application, the corresponding functional module in the present embodiment can be realized by corresponding hardware, it is also possible to by phase
The hardware that answers executes corresponding software and completes, and for example, aforesaid essential characteristic extraction module, can have
Execute mentioned by the statistics to discrete cosine transform coefficient histogram in bitmap to be estimated, extract described
Arbitrary frequency f of bitmap to be estimatedi,jCorresponding essential characteristic Ci,jHardware, such as essential characteristic extractor,
Can also be able to carry out corresponding computer program so as to complete the general processor of aforementioned function or other are hard
Part equipment;For another example aforesaid supplemental characteristic extraction module, can be by DCT coefficient in bitmap to be estimated
Histogrammic statistics simultaneously combines collimation technique, extracts arbitrary frequency f of the bitmap to be estimatedi,jCorresponding auxiliary
Feature C 'i,jHardware, for example supplemental characteristic extractor, or be able to carry out corresponding computer program so as to
(each embodiment that this specification is provided is all to complete the general processor of aforementioned function or other hardware devices
Foregoing description principle can be applied).
The essential characteristic extraction module 301 of 3 example of accompanying drawing can be united including first acquisition unit 401, first
Meter unit 402, the first interception unit 403, the first factor histogram feature vector acquiring unit 404 and first
Series unit 405, the JPEG compression of the estimation bitmap that the embodiment of the present invention four is provided as shown in Figure 4 quantify
The device of step-length, wherein:
First acquisition unit 401, for obtaining 64 frequencies of each 8 × 8 resolution ratio sub-block of bitmap to be estimated
The corresponding DCT coefficient of rate.
First statistic unit 402, the corresponding DCT coefficient of each frequency for counting 64 frequencies are absolute
The distribution of value, obtains corresponding 64 histograms of 64 frequencies.
First statistic unit 402 the step of realizing process and previous embodiment S1012 realize process class
Seemingly, with arbitrary frequency f of 64 frequenciesi,jCorresponding DCT coefficient for 0,1,0,15,3,1,
9,3,6,13,0,8,3,10,8,0,13,10,0,9,0,4,13,8,0,
6,1,10,6,4,8,0,15,16,0,12,4,4,9,13,7,10,
0,7,0,15,12,0 } as a example by, one of histogram such as accompanying drawing 2 that the first statistic unit 402 is obtained
Shown.
First interception unit 403, arbitrary in 64 histograms that the first statistic unit 402 is obtained for intercepting
Frequency fi,jL vertical bar in corresponding histogram, obtains arbitrary frequency f of bitmap to be estimatedi,jCorresponding vector
Gi,j, wherein, vectorial Gi,jFor L dimensional vector.
First factor histogram feature vector acquiring unit 404, for obtained to the first interception unit 403
Vectorial Gi,jEach DCT coefficient absolute value carry out Factorization, all positive factor that Statistics decomposition is obtained
The histogram of value, obtains arbitrary frequency f of bitmap to be estimatedi,jCorresponding factor histogram feature vector Fi,j,
Wherein, factor histogram feature vector Fi,jThe vectorial G that obtains of dimension and the first interception unit 403i,jDimension
Degree is identical.
First series unit 405, for joining the vectorial G that the first interception unit 403 is obtainedi,jStraight with the first factor
The factor histogram feature vector F that square figure characteristic vector acquiring unit 404 is obtainedi,j, the result [G of series connectioni,j,
Fi,j] be bitmap to be estimated arbitrary frequency fi,jCorresponding essential characteristic Ci,j.
The supplemental characteristic extraction module 302 of 3 example of accompanying drawing can include reference picture acquiring unit 501, the
Two acquiring units 502, the second statistic unit 503, the second interception unit 504, the second factor histogram feature
Vectorial acquiring unit 505 and the second series unit 506, as shown in Figure 5 the embodiment of the present invention five provide
Estimate the device of the JPEG compression quantization step of bitmap, wherein:
Reference picture acquiring unit 501, for obtaining the reference picture of bitmap to be estimated with collimation technique.
Reference picture acquiring unit 501 obtains the reference picture of bitmap to be estimated with collimation technique, and which is specifically real
The step of now see previous embodiment S1 is not repeated herein to step S3.
Second acquisition unit 502, for obtaining the every of reference picture that reference picture acquiring unit 501 is obtained
The corresponding DCT coefficient of 64 frequencies of individual 8 × 8 resolution ratio sub-block.
Second statistic unit 503, for the corresponding DCT coefficient absolute value of 64 each frequencies of frequency of statistics
Distribution, obtain corresponding 64 histograms of 64 frequencies.
Second interception unit 504, for intercepting arbitrary frequency f described in 64 histogramsi,jCorresponding
L vertical bar in histogram, obtains arbitrary frequency f of bitmap to be estimatedi,jCorresponding vector G 'i,j, wherein, to
Amount G 'i,jFor L dimensional vector.
Second factor histogram feature vector acquiring unit 505, for obtained to the second interception unit 504
Vectorial G 'i,jEach DCT coefficient absolute value carry out Factorization, all positive factor that Statistics decomposition is obtained
The histogram of value obtains arbitrary frequency fi,jCorresponding factor histogram feature vector F 'i'j, wherein, factor
Histogram feature vector F 'i,jThe vectorial G ' that obtains of dimension and the second interception unit 504i,jDimension identical.
Second series unit 506, for the vectorial G ' that the second interception unit 504 of connecting is obtainedi,jWith the second factor
The factor histogram feature vector F ' that histogram feature vector acquiring unit 505 is obtainedi'j, the result [G ' of series connectioni,j,
F′i,j] it is arbitrary frequency f of bitmap to be estimatedi,jCorresponding supplemental characteristic C 'i,j.
The device of the JPEG compression quantization step of the estimation bitmap of accompanying drawing 3 to 5 any example of accompanying drawing is acceptable
Module 601 is built including estimator, what the embodiment of the present invention six to eight was provided as shown in accompanying drawing 6-a to 6-c estimates
The device of the JPEG compression quantization step of meter bitmap.Estimator builds module 601 to be used for carrying in essential characteristic
Delivery block 301 extracts arbitrary frequency f of bitmap to be estimatedi,jCorresponding essential characteristic Ci,jBefore, build and quantify step
Long estimator.
The estimator of accompanying drawing 6-a to 6-c any example builds module 601 can include that training characteristics extract list
Unit 701 and training unit 702, what as shown in accompanying drawing 7-a to 7-c, the embodiment of the present invention nine to ten one was provided are estimated
The device of the JPEG compression quantization step of meter bitmap, wherein:
Training characteristics extraction unit 701, extracts position to be estimated for employing with essential characteristic extraction module 301
Scheme arbitrary frequency fi,jCorresponding essential characteristic Ci,jCi,jBitmap to be estimated is extracted with supplemental characteristic extraction module 302
Arbitrary frequency fi,jCorresponding supplemental characteristic C 'i,jIdentical mode, extracts arbitrary frequency f of bitmapi,jCorresponding training
Assemblage characteristic
Training unit 702, for arbitrary frequency f that is extracted with training characteristics extraction unit 701i,jCorresponding instruction
Practice assemblage characteristicAs the training characteristics of support vector machines, and using quantization step q as described group
Close featureTraining category, train the SVM to obtain arbitrary frequency f of bitmapi,jCorresponding quantization step
Long estimator.
It should be noted that the content such as information exchange between each module/unit of said apparatus, implementation procedure,
Due to being based on same design with the inventive method embodiment, the technique effect which brings is implemented with the inventive method
Example is identical, and particular content can be found in the narration in the inventive method embodiment, and here is omitted.
One of ordinary skill in the art will appreciate that all or part of step in the various methods of above-described embodiment
Can be by program to complete to instruct the hardware of correlation, the program can be stored in a computer-readable and deposit
In storage media, storage medium can include:Read-only storage (ROM, Read Only Memory), with
Machine access memory (RAM, Random Access Memory), disk or CD etc..
The method of the JPEG compression quantization step of the estimation bitmap for above embodiment of the present invention being provided and dress
Put and be described in detail, specific case used herein is carried out to the principle of the present invention and embodiment
Illustrate, the explanation of above example is only intended to help and understands the method for the present invention and its core concept;Meanwhile,
For one of ordinary skill in the art, according to the thought of the present invention, in specific embodiment and range of application
On all will change, in sum, this specification content should not be construed as limiting the invention.
Claims (10)
1. a kind of estimate bitmap JPEG compression quantization step method, it is characterised in that methods described includes:
By the statistics to discrete cosine transform coefficient histogram in bitmap to be estimated, extract described in wait to estimate
Arbitrary frequency f of meter bitmapi,jCorresponding essential characteristic Ci,j;
Collimation technique is counted and is combined by histogrammic to DCT coefficient in the bitmap to be estimated, extract institute
State arbitrary frequency f of bitmap to be estimatedi,jCorresponding supplemental characteristic C 'i,j, the supplemental characteristic C 'i,jIt is and the base
Eigen Ci,jDimension identical vector;
Connect essential characteristic Ci,jWith the supplemental characteristic C 'i,j, obtain assemblage characteristic [Ci,j, C 'i,j];
All described assemblage characteristic [C by the bitmap to be estimatedi,j, C 'i,j] input quantization step-size estimation device, obtain
Obtain the estimate of the JPEG compression quantization step of the bitmap to be estimated.
2. as claimed in claim 1 estimate bitmap JPEG compression quantization step method, it is characterised in that
The statistics by discrete cosine transform coefficient histogram in bitmap to be estimated, extract described in wait to estimate
Arbitrary frequency f of meter bitmapi,jCorresponding essential characteristic Ci,j, including:
Obtain the corresponding DCT coefficient of 64 frequencies of each 8 × 8 resolution ratio sub-block of the bitmap to be estimated;
The distribution of the corresponding DCT coefficient absolute value of each frequency of 64 frequencies is counted, is obtained described
Corresponding 64 histograms of 64 frequencies;
Intercept arbitrary frequency f described in 64 histogramsi,jIn corresponding histogram, L vertical bar, obtains
Arbitrary frequency fi,jCorresponding vector Gi,j, the vector Gi,jFor L dimensional vector;
To the vector Gi,jEach DCT coefficient absolute value carry out Factorization, Statistics decomposition is obtained
The histogram of all positive factor value, obtains arbitrary frequency fi,jCorresponding factor histogram feature vector Fi,j,
The factor histogram feature vector Fi,jDimension and the vector Gi,jDimension identical;
The series connection vector Gi,jWith factor histogram feature vector Fi,j, the result [G of the series connectioni,j, Fi,j]
For arbitrary frequency f of the bitmap to be estimatedi,jCorresponding essential characteristic Ci,j.
3. as claimed in claim 1 estimate bitmap JPEG compression quantization step method, it is characterised in that
Described by statistics histogrammic to DCT coefficient in the bitmap to be estimated and combine collimation technique, extract institute
State arbitrary frequency f of bitmap to be estimatedi,jCorresponding supplemental characteristic C 'i,j, including:
The reference picture of the bitmap to be estimated is obtained with collimation technique;
Obtain the corresponding DCT coefficient of 64 frequencies of each 8 × 8 resolution ratio sub-block of the reference picture;
The distribution of the corresponding DCT coefficient absolute value of described each frequency of 64 frequencies is counted, obtains described 64
Corresponding 64 histograms of individual frequency;
Intercept arbitrary frequency f described in 64 histogramsi,jIn corresponding histogram, L vertical bar, obtains
Arbitrary frequency fi,jCorresponding vector G 'i,j, the vector G 'i,jFor L dimensional vector;
To the vector G 'i,jEach DCT coefficient absolute value carry out Factorization, Statistics decomposition is obtained
The histogram of all positive factor value obtains arbitrary frequency fi,jCorresponding factor histogram feature vector F 'i,j,
The factor histogram feature vector F 'i,jDimension and the vector G 'i,jDimension identical;
The series connection vector G 'i,jWith factor histogram feature vector F 'i,j, the result [G ' of the series connectioni,j, F 'i,j]
For arbitrary frequency f of the bitmap to be estimatedi,jCorresponding supplemental characteristic C 'i,j.
4. as described in claims 1 to 3 any one estimation bitmap JPEG compression quantization step side
Method, it is characterised in that the system by discrete cosine transform coefficient histogram in bitmap to be estimated
Meter, extracts arbitrary frequency f of the bitmap to be estimatedi,jCorresponding essential characteristic Ci,jBefore, methods described is also wrapped
Include:
Build the quantization step estimator.
5. the method for the JPEG compression quantization step for estimating bitmap as claimed in claim 4, its feature exists
In, described build the quantization step estimator, including:
Using with extract arbitrary frequency f of the bitmap to be estimatedi,jCorresponding essential characteristic Ci,jCi,jSpecial with auxiliary
Levy C 'i,jIdentical mode, extracts arbitrary frequency f of bitmapi,jCorresponding training assemblage characteristic
With arbitrary frequency fi,jCorresponding training assemblage characteristicTraining as support vector machines
Feature, and using quantization step q as the assemblage characteristicTraining category, train the SVM with
To arbitrary frequency fi,jCorresponding quantization step estimator.
6. a kind of estimate bitmap JPEG compression quantization step device, it is characterised in that described device bag
Include:
Essential characteristic extraction module, for by discrete cosine transform coefficient Nogata in bitmap to be estimated
The statistics of figure, extracts arbitrary frequency f of the bitmap to be estimatedi,jCorresponding essential characteristic Ci,j;
Supplemental characteristic extraction module, for by the histogrammic statistics of DCT coefficient in the bitmap to be estimated
And collimation technique is combined, extract arbitrary frequency f of the bitmap to be estimatedi,jCorresponding supplemental characteristic C 'i,j, institute
State supplemental characteristic C 'i,jIt is and essential characteristic Ci,jDimension identical vector;
Assemblage characteristic acquisition module, for essential characteristic C of connectingi,jWith the supplemental characteristic C 'i,j, obtain
Assemblage characteristic [Ci,j, C 'i,j];
Estimate acquisition module, for by all described assemblage characteristic [C of the bitmap to be estimatedi,j, C 'i,j]
Input quantifies step-size estimation device, obtains the estimate of the JPEG compression quantization step of the bitmap to be estimated.
7. as claimed in claim 6 estimate bitmap JPEG compression quantization step device, it is characterised in that
The essential characteristic extraction module includes:
First acquisition unit, for obtaining 64 frequencies of each 8 × 8 resolution ratio sub-block of the bitmap to be estimated
The corresponding DCT coefficient of rate;
First statistic unit, the corresponding DCT coefficient of each frequency for counting 64 frequencies are absolute
The distribution of value, obtains corresponding 64 histograms of 64 frequencies;
First interception unit, for intercepting arbitrary frequency f described in 64 histogramsi,jCorresponding Nogata
In figure L vertical bar, obtains arbitrary frequency fi,jCorresponding vector Gi,j, the vector Gi,jFor L dimensional vector;
First factor histogram feature vector acquiring unit, for the vector Gi,jEach DCT system
Number absolute value carries out Factorization, the histogram of all positive factor value that Statistics decomposition is obtained, and obtains described appointing
One frequency fi,jCorresponding factor histogram feature vector Fi,j, the factor histogram feature vector Fi,jDimension
With the vector Gi,jDimension identical;
First series unit, for joining the vector Gi,jWith factor histogram feature vector Fi,j, the series connection
Result [Gi,j, Fi,j] it is arbitrary frequency f of the bitmap to be estimatedi,jCorresponding essential characteristic Ci,j.
8. as claimed in claim 6 estimate bitmap JPEG compression quantization step device, it is characterised in that
The supplemental characteristic extraction module includes:
Reference picture acquiring unit, for obtaining the reference picture of the bitmap to be estimated with collimation technique;
Second acquisition unit, for obtaining 64 frequencies of each 8 × 8 resolution ratio sub-block of the reference picture
Corresponding DCT coefficient;
Second statistic unit, for counting the corresponding DCT coefficient absolute value of described each frequency of 64 frequencies
Distribution, obtain corresponding 64 histograms of 64 frequencies;
Second interception unit, for intercepting arbitrary frequency f described in 64 histogramsi,jCorresponding Nogata
In figure L vertical bar, obtains arbitrary frequency fi,jCorresponding vector G 'i,j, the vector G 'i,jFor L dimensional vector;
Second factor histogram feature vector acquiring unit, for the vector G 'i,jEach DCT system
Number absolute values carry out Factorization, and the histogram of all positive factor value that Statistics decomposition is obtained obtains described arbitrary
Frequency fi,jCorresponding factor histogram feature vector F 'i,j, the factor histogram feature vector F 'i,jDimension with
The vector G 'i,jDimension identical;
Second series unit, for the vector G ' that connectsi,jWith factor histogram feature vector F 'i,j, the string
Result [the G ' of connectioni,j, F 'i,j] it is arbitrary frequency f of the bitmap to be estimatedi,jCorresponding supplemental characteristic C 'i,j.
9. as described in claim 6 to 8 any one estimation bitmap JPEG compression quantization step dress
Put, it is characterised in that described device also includes:
Estimator builds module, arbitrary for extracting the bitmap to be estimated in the essential characteristic extraction module
Frequency fi,jCorresponding essential characteristic Ci,jBefore, the quantization step estimator is built.
10. the device of the JPEG compression quantization step for estimating bitmap as claimed in claim 9, its feature exists
In the estimator builds module to be included:
Training characteristics extraction unit, for adopting and extracting arbitrary frequency f of the bitmap to be estimatedi,jCorresponding base
Eigen Ci,jCi,jWith supplemental characteristic C 'i,jIdentical mode, extracts arbitrary frequency f of bitmapi,jCorresponding training combination
Feature
Training unit, for arbitrary frequency fi,jCorresponding training assemblage characteristicAs supporting vector
The training characteristics of machine SVM, and using quantization step q as the assemblage characteristicTraining category, training
The SVM is to obtain arbitrary frequency fi,jCorresponding quantization step estimator.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510531953.5A CN106485738B (en) | 2015-08-26 | 2015-08-26 | A kind of method and apparatus of JPEG compression quantization step that estimating bitmap |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510531953.5A CN106485738B (en) | 2015-08-26 | 2015-08-26 | A kind of method and apparatus of JPEG compression quantization step that estimating bitmap |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106485738A true CN106485738A (en) | 2017-03-08 |
CN106485738B CN106485738B (en) | 2019-04-05 |
Family
ID=58234455
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510531953.5A Active CN106485738B (en) | 2015-08-26 | 2015-08-26 | A kind of method and apparatus of JPEG compression quantization step that estimating bitmap |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106485738B (en) |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102413328A (en) * | 2011-11-11 | 2012-04-11 | 中国科学院深圳先进技术研究院 | Double compression detection method and system of joint photographic experts group (JPEG) image |
CN102595138A (en) * | 2012-02-29 | 2012-07-18 | 北京大学 | Method, device and terminal for image compression |
CN103067713A (en) * | 2013-01-06 | 2013-04-24 | 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 |
CN104486524A (en) * | 2014-12-30 | 2015-04-01 | 中国科学院深圳先进技术研究院 | Method for detecting whether images are subjected to two times of JPEG compression with same compression quality |
-
2015
- 2015-08-26 CN CN201510531953.5A patent/CN106485738B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102413328A (en) * | 2011-11-11 | 2012-04-11 | 中国科学院深圳先进技术研究院 | Double compression detection method and system of joint photographic experts group (JPEG) image |
CN102595138A (en) * | 2012-02-29 | 2012-07-18 | 北京大学 | Method, device and terminal for image compression |
CN103067713A (en) * | 2013-01-06 | 2013-04-24 | 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 |
CN104486524A (en) * | 2014-12-30 | 2015-04-01 | 中国科学院深圳先进技术研究院 | Method for detecting whether images are subjected to two times of JPEG compression with same compression quality |
Also Published As
Publication number | Publication date |
---|---|
CN106485738B (en) | 2019-04-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Chierchia et al. | Epigraphical projection and proximal tools for solving constrained convex optimization problems | |
CN107911698B (en) | The embedding grammar and detection method of video digital watermark information | |
CN106713964A (en) | Method of generating video abstract viewpoint graph and apparatus thereof | |
JP2021509747A (en) | Hardware-based pooling system and method | |
CN110263909A (en) | Image-recognizing method and device | |
JP2013529814A5 (en) | ||
CN102630011A (en) | Compressive perceptual coding and decoding method and system in video sensor network | |
MY152525A (en) | Video abstraction | |
CN105469353B (en) | The embedding grammar and device and extracting method and device of watermarking images | |
US20170200307A1 (en) | Constructing a 3d structure | |
Grogan et al. | L2 registration for colour transfer | |
ITTO20130835A1 (en) | PROCEDURE FOR PRODUCING COMPACT DESCRIBERS FROM POINTS OF INTEREST OF DIGITAL IMAGES, SYSTEM, EQUIPMENT AND CORRESPONDENT COMPUTER PRODUCT | |
CN103745443B (en) | The method and apparatus for improving picture quality | |
CN109286819A (en) | Combine explicit image encryption, decryption method and the device of compression | |
Li et al. | Fast principal component analysis for hyperspectral imaging based on cloud computing | |
CN107924576A (en) | Video image for video stabilization aligns | |
US20240202886A1 (en) | Video processing method and apparatus, device, storage medium, and program product | |
Junayed et al. | HiMODE: A hybrid monocular omnidirectional depth estimation model | |
CN103150709A (en) | Quaternion field colored image compressed sensing recovery method based on Quasi Newton algorithm | |
CN104683818B (en) | Method for compressing image based on biorthogonal invariant set m ultiwavelet | |
CN109544557A (en) | Principal component analysis conversion method and device based on block | |
CN106485738A (en) | A kind of method and apparatus of the JPEG compression quantization step for estimating bitmap | |
CN104243986B (en) | Compression video acquisition and reconfiguration system based on data-driven tensor subspace | |
CN107181957A (en) | A kind of video watermark source tracing method based on hadoop platform architectures | |
CN106488250A (en) | A kind of method and apparatus of the dual compressed first pressure quantization step of estimation JPEG |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |