CN108366295A - Visual classification feature extracting method, transcoding weight contracting detection method and storage medium - Google Patents
Visual classification feature extracting method, transcoding weight contracting detection method and storage medium Download PDFInfo
- Publication number
- CN108366295A CN108366295A CN201810144370.0A CN201810144370A CN108366295A CN 108366295 A CN108366295 A CN 108366295A CN 201810144370 A CN201810144370 A CN 201810144370A CN 108366295 A CN108366295 A CN 108366295A
- Authority
- CN
- China
- Prior art keywords
- video
- frame
- classified types
- weight contracting
- pixels
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
- H04N21/43—Processing of content or additional data, e.g. demultiplexing additional data from a digital video stream; Elementary client operations, e.g. monitoring of home network or synchronising decoder's clock; Client middleware
- H04N21/44—Processing of video elementary streams, e.g. splicing a video clip retrieved from local storage with an incoming video stream, rendering scenes according to MPEG-4 scene graphs
- H04N21/4402—Processing of video elementary streams, e.g. splicing a video clip retrieved from local storage with an incoming video stream, rendering scenes according to MPEG-4 scene graphs involving reformatting operations of video signals for household redistribution, storage or real-time display
- H04N21/440218—Processing of video elementary streams, e.g. splicing a video clip retrieved from local storage with an incoming video stream, rendering scenes according to MPEG-4 scene graphs involving reformatting operations of video signals for household redistribution, storage or real-time display by transcoding between formats or standards, e.g. from MPEG-2 to MPEG-4
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/17—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object
- H04N19/172—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being an image region, e.g. an object the region being a picture, frame or field
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/177—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a group of pictures [GOP]
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/70—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by syntax aspects related to video coding, e.g. related to compression standards
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N21/00—Selective content distribution, e.g. interactive television or video on demand [VOD]
- H04N21/40—Client devices specifically adapted for the reception of or interaction with content, e.g. set-top-box [STB]; Operations thereof
- H04N21/43—Processing of content or additional data, e.g. demultiplexing additional data from a digital video stream; Elementary client operations, e.g. monitoring of home network or synchronising decoder's clock; Client middleware
- H04N21/44—Processing of video elementary streams, e.g. splicing a video clip retrieved from local storage with an incoming video stream, rendering scenes according to MPEG-4 scene graphs
- H04N21/44008—Processing of video elementary streams, e.g. splicing a video clip retrieved from local storage with an incoming video stream, rendering scenes according to MPEG-4 scene graphs involving operations for analysing video streams, e.g. detecting features or characteristics in the video stream
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The present invention provides a kind of visual classification feature extracting method, transcoding weight contracting detection method and storage medium, the visual classification feature extracting method includes:The PU classified types of video frame are extracted using Visual Analyzer, and the PU classified types for the video frame extracted are marked by basic unit of block of pixels;Count the corresponding block of pixels number of each PU classified types of first P frame in each group of continuous pictures in video;By the corresponding block of pixels number averaged of each PU classified types of first P frame in each group of continuous pictures, the characteristic of division of each PU classified types of first P frame in all groups of continuous pictures is obtained.The present invention uses the less Dimension Characteristics of number, has reached the verification and measurement ratio of higher weight contracting video.
Description
Technical field
The present invention relates to technical field of video processing, more particularly to a kind of visual classification feature extracting method, video code conversion
Weight contracting detection method and computer readable storage medium.
Background technology
Under the fast development of internet instantly, the acquisition and transmission of digital video become increasingly mature and universal.Meanwhile
Have when the case where Video editing software that function is become stronger day by day also increasingly is pursued by the common people, this makes digital video be tampered
Occur.When the video being tampered is for industries such as the administration of justice, media, can distort the facts truth, can then cause judicial erroneous judgement, matchmaker
The unfounded serious consequence of body.Then, the authenticity and integrity of digital video urgently solves the problems, such as society need instantly.
Currently, in terms of research both domestic and external is concentrated mainly on digital picture, a large amount of achievement in research is also obtained.For example,
The illegal copies of drm image are detected, the duplication movement in image detects, and the image and photographs that computer is generated
It distinguishes.Due to the containing much information of video, the features such as mode is various is distorted, causes video evidence obtaining research difficult.It relies on
Image forensics progress of research, in recent years video forensic technologies also obtain tremendous development.
Common video, which is distorted means generally and need to be undergone decoding, deletes frame or be inserted into frame etc., distorts operation, and what is be tampered regards
Frequency sequence is required for just regenerating video code flow by compressing again.Therefore, whether detection video can be used as by weight contracting
A kind of technological means whether detection video is tampered.In the prior art, there are many ways to counterweight compressed detected, such as:With
When carrying out weight contracting to MPEG video with bit rate, than double times pressures of DCT coefficient variable number of single compression and double second compressions
The variable number of contracting and three second compressions is more, and the heavy compressed detected of video is carried out using this phenomenon.By checking weight contracting video
Block effect intensity rule and its mean value variation come detect MPEG video weight contracting.The motion compensation of adjacent P frames is utilized
The difference of edge effect, and by judging to contract to detect the weight of MPEG video with the presence or absence of spike in Fourier transform domain.Profit
MPEG-2 singles compression video is distinguished with the convex characteristic of quantization DCT coefficient statistic histogram in weight contracting video and weight contracting regards
Frequently, algorithm is suitable for detecting is contracted using the weight of different MPEG-2 encoders, distorts frame deletion with robustness.One kind is with non-
The detection algorithm that the probability of zero quantization AC coefficients is characterized compresses video and dual compression video for distinguishing H.264 single, when
Second of compression quantization parameter achieves very high classification accuracy when being less than compression quantization parameter for the first time.A kind of identical quantization ginseng
H.264 video Multiple Compression evidence obtaining algorithm under several, is constructed with the ratio difference of different quantization DCT coefficients between adjacent three second compression
Feature set containing quartile realizes as the input of support vector machines and compresses video and Multiple Compression video to single
Classification.Carried algorithm has high-class precision, has stronger robustness to copy/paste attack and frame deletion attack.
As newest video encoding standard, HEVC has attracted the concern of considerable researcher.H.264 phase
Than, in the case of same video quality, in the case of HEVC provides double data compression ratio namely identical bit rate, HEVC energy
Greatly improve video quality.The high resolution that it is supported is up to 8192 × 4320, wherein also including 8k UHD.Therefore, the prior art
In propose for HEVC video weight contract plus survey method, such as:One kind combining statistics spy based on adjacent DCT coefficient to odd even
The HEVC weight contracting Video Detection Algorithms of property.Propose in the prior art simultaneously co-occurrence matrix based on quantization DCT coefficient and
HEVC video weight contracting detection algorithms based on Markov characteristic optimizations.It is divided using each PU of I frames in HEVC weight contracting videos
The corresponding block number purpose jumping phenomenon of type, it is proposed that HEVC videos weight contracting evidence obtaining algorithm under different bit rates.But it is existing
It is more for the characteristic of division dimension of the method for HEVC videos weight compressed detected in technology, it is computationally intensive, do not reach higher yet
Verification and measurement ratio.
Therefore, in order to solve the above-mentioned technical problem, need a kind of characteristic of division dimension few, the one kind for reaching high detection rate regards
Frequency division category feature extracting method and video code conversion weight contracting detection method.
Invention content
The purpose of the present invention is to provide a kind of visual classification feature extracting method and video code conversion weight contracting detection method,
To solve at least one defect in the prior art.
An aspect of of the present present invention provides a kind of and video code conversion weight contracting detection method, the method includes:
Predicting unit (PU) classified types of video frame, and the video that will be extracted are extracted using Visual Analyzer
The PU classified types of frame are marked by basic unit of block of pixels;
Count the corresponding block of pixels number of each PU classified types of first P frame in each group of continuous pictures in video;
The corresponding block of pixels number of each PU classified types of first P frame in each group of continuous pictures is sought average
Value, obtains the characteristic of division of each PU classified types of first P frame in all groups of continuous pictures.
Preferably, when the PU classified types extraction of the video frame, the RGB of the Visual Analyzer border color is set
Component, preferably could be provided as (255,0,255) or other can distinguish PU and divide any suitable of frame and video content
Component value.
Preferably, the PU classified types of the video frame are marked using N × N block of pixels as basic unit, and wherein N is 4
Or 4 integer multiple.Such as it can be marked using 8 × 8 block of pixels as basic unit.
Preferably, in each group of continuous pictures first P frame the corresponding block of pixels number of each PU classified types
Averaged is realized by following formula:
Wherein Pi={ pi,0,pi,1..., pi,24(i=1,2 ..., M), M be video in include it is continuous
The group number of picture.
Another aspect of the present invention is to provide a kind of video code conversion weight contracting detection method, including:
It randomly selects the same number of single compression video and weight contracting video is sent into support vector machines as training sample;
Video is compressed to the single as follows and weight contracting video carries out visual classification feature extraction:PU is drawn
Classify type analysis, the PU classified types of video frame is extracted using Visual Analyzer, and by the PU for the video frame extracted
Classified types are marked by basic unit of block of pixels;Count in video in each group of continuous pictures each of first P frame
The corresponding block of pixels number of PU classified types;Each PU classified types of first P frame in each group of continuous pictures are corresponding
Block of pixels number averaged obtains the characteristic of division of each PU classified types of first P frame in all groups of continuous pictures;
The support vector machines builds decision function according to the characteristic of division of extraction;
Randomly select for the single of test compress video and weight contracting video as test sample feeding it is described support to
Amount machine, the video that the support vector machines exports discriminating test according to the decision function are that single compresses video or weight contracts
The classification and Detection result of video.
Preferably, the method further includes calculating the evaluation index for indicating classification performance in the following way:Wherein, AR is evaluation index, and TNR is the ratio for being determined as single compression video;TPR is that judgement is attached most importance to
Compress the ratio of video.
Preferably, the method further includes:The verification and measurement ratio for indicating video compress detection is calculated in the following way:Wherein, n is the number of test and the different video sample of training.
The weight contracting video be original video with the first bit rate carry out H.261, H.263, H.263+, H.264,
Any one of MPEG-1, MPEG-2 and MPEG-4 reference format are compressed.Such as the weight contracting video is original video
It is H.264 compressed with the first bit rate, carrying out HEVC to decoded video with the second bit rate again after decoded compresses to obtain
Video.
Preferably, single compression video is the video that original video carries out that HEVC is compressed with the second bit rate.
The PU classified types of the video frame are marked using N × N block of pixels as basic unit, and wherein N is 4 or 4
Integer multiple.For example, when the PU classified types extraction of the video frame, the RGB of the Visual Analyzer border color is chosen
Component is (255,0,255);The PU classified types of the video frame are marked using 8 × 8 block of pixels as basic unit.
Preferably, in each group of continuous pictures first P frame the corresponding block of pixels number of each PU classified types
Averaged is realized by the following method:
Wherein Pi={ pi,0,pi,1..., pi,24(i=1,2 ..., M), M be video in include it is continuous
The group number of picture.
Another aspect of the present invention also provides a kind of computer readable storage medium, is deposited in the computer readable storage medium
Computer program is contained, is performed in the computer program and realizes method and step as described above.
A kind of visual classification feature extracting method and video code conversion weight contracting detection method provided by the invention, for
H.264 weight compressed detected is carried out to the video code conversion weight contracting video of HEVC standard, the characteristic of division dimension of extraction is few, Neng Gouda
To higher verification and measurement ratio.In addition, method provided by the invention be equally readily applicable to detection first time video encoding standard be
Video encoding standard before other HEVC standards, for example, be readily applicable to H.261, H.263, H.263+, MPEG-1,
Any one of MPEG-2 and MPEG-4 standard.
It should be appreciated that aforementioned description substantially and follow-up description in detail are exemplary illustration and explanation, it should not
As the limitation to the claimed content of the present invention.
Description of the drawings
With reference to the attached drawing of accompanying, the more purposes of the present invention, function and advantage are by the as follows of embodiment through the invention
Description is illustrated, wherein:
Fig. 1 is the schematic diagram of PU dividing modes under different prediction modes;
Fig. 2 is the flow diagram of visual classification feature extracting method of the present invention;
Fig. 3 a to Fig. 3 d are the P frame PU classified types schematic diagrames of single compression video and weight contracting video;
Fig. 4 is the label schematic diagram of PU classified types of the present invention;
Fig. 5 is the flow diagram of video code conversion weight contracting detection method of the present invention.
Specific implementation mode
By reference to exemplary embodiment, the purpose of the present invention and function and the side for realizing these purposes and function
Method will be illustrated.However, the present invention is not limited to exemplary embodiment as disclosed below;Can by different form come
It is realized.The essence of specification is only to aid in the detail of the various equivalent modifications Integrated Understanding present invention.
Hereinafter, the embodiment of the present invention will be described with reference to the drawings.In the accompanying drawings, identical reference numeral represents identical
Or similar component or same or like step.Present disclosure is said below by specific embodiment
Bright, video interpolater needs other video compression format to video sequence after carrying out frame deletion, insertion etc. to video and distorting
Carry out weight contracting.Compared with other coding standards, H.264 coding encodes mostly concerned with HEVC, and coding scheme is substantially similar.It usurps
After the person of changing distorts the original video of H.264 format, re-compressed using HEVC codings when compressing again.It is used in this example
H.264 standard of the coding standard as first time video compress.
Video is after the contracting of transcoding weight, due to the change of key frame pixel value, inner first P of one group of continuous pictures (GOP)
Frame is using key frame as when carrying out inter prediction encoding with reference to frame, and PU classified types also can correspondingly change in frame.
Based on this phenomenon, the present invention is made using the corresponding block of pixels number of each PU classified types in the such frame of statistics with histogram with this
The transcoding weight contracting of video is detected for characteristic of division.
It is more clear to the transcoding weight compressed detected of HEVC in order to enable the present invention to be directed to other coding standards in embodiment
Clear explaination, it is necessary to which P frame PU classified types are illustrated.HEVC codings use the hybrid coding frame of prediction plus transformation
Frame.Three basic units, i.e. coding unit (CU), predicting unit (PU) and converter unit (TU) are introduced in HEVC codings,
So that the coding mode of HEVC ratios H.264/AVC is more flexible.Wherein, CU is coding basic unit, PU in frame and
Inter-prediction, TU are used for transform and quantization.PU is the basic unit for containing predictive information, a CU can be divided into one or
Multiple PU, PU prediction mode, which can be divided into, to be skipped, in frame and interframe.The signal of PU classified types under different mode as shown in Figure 1
Figure, when prediction mode is skip mode a, PU sizes are only 2N × 2N;When prediction mode is frame mode b, PU has
Two kinds of partition modes of 2N × 2N and N × N;When prediction mode is inter-frame mode c, PU has 8 kinds of partition modes, including 2N × 2N and
N × N two kinds of square segmentation pattern c1,2N × N and N × 2N two symmetry division patterns c2 and 2N × nU, 2N × nD, nL ×
The 2N and asymmetric Fractionation regimen c3 of nR × 2N tetra-.Wherein asymmetric Fractionation regimen c3 is optional mode, can be configured by encoding
Dependent parser in file controls it and is turned on and off.
In transcoded video weight compression process, the predictive information that PU blocks include illustrates the prediction process of CU blocks.Such as Fig. 2 institutes
It is shown as the flow diagram of visual classification feature extracting method of the present invention, as shown in Fig. 2, a kind of video provided in an embodiment of the present invention
Characteristic of division extracting method includes the following steps:
Step S101 analyzes PU classified types, extracts the PU classified types of video frame and be marked.
PU classified types are analyzed, single compression video is shown such as Fig. 3 a to Fig. 3 d and are H.264 regarded to the contracting of HEVC weight
The P frame PU classified types schematic diagrames of frequency by analyzing PU classified types in embodiment, and then extract the characteristic of division of video.
Fig. 3 a are to be encoded using HEVC, with inner first P of first group of continuous pictures (GOP) of the bridge_far of 3.5M Bit-Rate Reductions
The PU classified types of frame.Fig. 3 b are to be encoded with 3.5M bit rate pressures using H.264 encoding with after 3M Bit-Rate Reductions, then with HEVC
The PU classified types of the inner first P frame of first group of continuous pictures (GOP) after contracting.Fig. 3 c are to use H.264 to encode with 3.5M ratios
After special rate compression, then with HEVC codings with first group of continuous pictures (GOP) inner first of 3.5M Bit-Rate Reductions bridge_far
The PU classified types of a P frames.Fig. 3 d are to be encoded with 3.5M bits using H.264 encoding with after 4M Bit-Rate Reductions, then with HEVC
Rate compresses the PU classified types of the inner first P frame of first group of continuous pictures (GOP) of bridge_far.
The corresponding block of pixels number of 1. each PU classified types of table
As can be seen from Table 1, the corresponding block of pixels number of each PU classified types of 3M-3.5M, 3.5M-3.5M and 4M-3.5
Distribution trend is roughly the same, and block of pixels number distribution difference corresponding with the PU classified types of 3.5M is larger.Especially PU is divided
When type is 4 × 4,8 × 8,4 × 8,8 × 4, the corresponding pixel of each PU classified types of 3M-3.5M, 3.5M-3.5M and 4M-3.5
Block number mesh block of pixels number distribution corresponding with the PU classified types of 3.5M has larger difference.
The analysis of PU classified types and table 1 are shown from the above analysis, after H.264 encoding, then with after HEVC coding compressions
The inner first P frame of first group of continuous pictures (GOP) PU classified types based on small pixel block;HEVC coding pressures are intended for single use
The PU classified types of the inner first P frame of first group of continuous pictures (GOP) after contracting are based on big block of pixels.It uses as a result, H.264
After coding, then it is that fine PU divides class to encode the PU of the compressed inner first P frame of first group of continuous pictures (GOP) with HEVC
Type.
It should be appreciated that in H.264 coding compression, dct transform, the phase between block and block are carried out to each block of pixels respectively
Closing property is ignored.There is discontinuous saltus step in the pixel value of the boundary of block and block.When video is encoded with HEVC again to be compressed, due to
The boundary of the influence of blocking artifact, block and block just needs smaller PU classified types to express image saltus step herein.Except H.264
Except other HEVC before video encoding standard also use block DCT transform, equally exist blocking artifact, it can thus be anticipated that
When the video encoding standard before HEVC being used to carry out video compress and then transcoding as HEVC format videos, can all it need more
Small PU classified types express the image saltus step at block boundary.
According to the present invention, after embodiment is by analyzing PU classified types, video frame is extracted using Visual Analyzer
PU classified types, and the PU classified types for the video frame extracted are marked by basic unit of block of pixels.
Specifically, in embodiment Visual Analyzer may be used such as Gitl_HEVC_Analyze video analysis it is soft
Part or other any suitable softwares carry out video frame the extraction of PU classified types.In order to which video background and PU are divided class
Type boundary distinguishes, and the RGB component of Visual Analyzer border color is set as (255,0,255).
According to the present invention, the PU classified types of video frame are marked using 8 × 8 block of pixels as basic unit in embodiment,
PU classified types are marked in a manner of label in embodiment, table 2 is the corresponding PU classified types of label, according to this hair
It is bright, totally 25 kinds of PU classified types in embodiment.The label schematic diagram of PU classified types of the present invention as shown in Figure 4.
The corresponding PU classified types of 2. label of table
Step S102 counts the corresponding picture of each PU classified types of first P frame in each group of continuous pictures in video
Plain block number mesh.
Count the corresponding block of pixels number of each PU classified types of first P frame in each group of continuous pictures in video.
The corresponding block of pixels number of each PU classified types of first P frame in each group of continuous pictures is denoted as:Pi={ pi,0,
pi,1..., pi,24(i=1,2 ..., M), M is the group number for the continuous pictures for including, i.e., each P in videoiIn have recorded 25 kinds
Corresponding 8 × 8 block of pixels number of PU classified types.
Step S103 extracts the characteristic of division of video.
The corresponding block of pixels number of each PU classified types of first P frame in each group of continuous pictures is sought average
Value, obtains the characteristic of division of each PU classified types of first P frame in all groups of continuous pictures.In each group of continuous pictures
The corresponding block of pixels number averaged of each PU classified types of one P frame is realized by following formula:
Wherein Pi={ pi,0,pi,1..., pi,24(i=1,2 ..., M), M be video in include it is continuous
The group number of picture.
Each group is continuously continued the average value of the corresponding block of pixels number of each PU classified types of first P frame in picture
The most each group histogram for continuously continuing the corresponding block of pixels number of each PU classified types of first P frame in picture, obtains
The characteristic of division of video.
A kind of extracting method of visual classification feature is to video code conversion weight compressed detected through the invention, as shown in Figure 5 originally
The flow diagram of invention video code conversion weight contracting detection method, specifically a kind of video code conversion weight contracting detection method include:
Step S201, randomly selects the same number of single compression video and weight contracting video is sent into branch as training sample
Hold vector machine.
Weight contracting video be original video H.264 compressed with the first bit rate, it is decoded after again to decoded video
The video that HEVC compresses is carried out with the second bit rate.Single compression video is that original video carries out HEVC pressures with the second bit rate
Contract obtained video.
Specifically, single compression video and weight contracting video are made in embodiment first as detection target.
Using 34 unpressed YUV sequences as initial video, including 17 QCIF format videos, (resolution ratio is
176 × 144) and 17 CIF format videos (resolution ratio is 352 × 288).In order to increase sample size, each video is divided
At the non-overlapping video clip that length is 100 frames.Finally, 36 QCIF video clips and 43 CIF video clips are collectively generated.
HM10.0 carries out HEVC using encoder_lowdelay_P_main configuration files and codes and decodes process.JM is adopted
Process is H.264 coded and decoded with encoder_main configuration files.Frame per second, I frame periods and GOP sizes are respectively set to
30,4 and 4.
It makes single and compresses video:To original video with the second bit rate (B2) carry out HEVC compress to obtain.
Make weight contracting video:To original video with the first bit rate (B1) H.264 compressed, it is decoded after again to solution
Code rear video is with the second bit rate (B2) carry out HEVC compressions.
Since QCIF and CIF videos have different spatial resolutions in embodiment, different bit rates should be selected
To ensure the visual quality of encoded video.For QCIF videos, the first bit rate (B1) and the second bit rate (B2) value respectively from
{ 100,200,300 } selection in (kbps) and { 200,300,400 } (kbps).For CIF videos, the first bit rate (B1) and the
Two bit rate (B2) selection from { 3,3.5,4 } (Mbps) and { 3.5,4,4.5 } (Mbps) respectively.
In the single of above-mentioned making compresses video and weight contracting video, the same number of single compression video is chosen immediately
Support vector machines (SVM) is sent into weight contracting video as training sample to be trained.The present embodiment, for regarding for QCIF formats
Frequency randomly chooses 30 single compression videos and 30 weight contracting videos are trained.Video random selection for CIF formats
35 single compression videos and 35 weight contracting videos are trained.
Training stage, support vector machines (SVM) executes step 2 and step 3 builds decision function.
Step 202 compresses video and weight contracting video extraction characteristic of division to single.
According to the present invention, embodiment compresses video to single as follows and weight contracting video carries out visual classification spy
Sign extraction:
The PU classified types of video frame are extracted using Visual Analyzer, and the PU for the video frame extracted is divided
Type is marked by basic unit of block of pixels;
Count the corresponding block of pixels number of each PU classified types of first P frame in each group of continuous pictures in video;
The corresponding block of pixels number of each PU classified types of first P frame in each group of continuous pictures is sought average
Value, obtains the characteristic of division of each PU classified types of first P frame in all groups of continuous pictures.
Step S203, support vector machines builds decision function according to the characteristic of division of extraction.
The extraction of visual classification feature in step 2 is hereinbefore had been presented for illustrating in detail, which is not described herein again.
Preferably, in embodiment support vector machines can choose with the LIBSVM open source softwares of SVMcg kernels or its
He has the software of similar functions as grader.
Step S204, it randomly selects and compresses video and weight contracting video as test sample feeding institute for the single of test
State support vector machines, output category result.
Randomly select for the single of test compress video and weight contracting video as test sample feeding it is described support to
Amount machine, the video that the support vector machines exports discriminating test according to the decision function are that single compresses video or weight contracts
The classification and Detection result of video.
According to the present invention, the present embodiment can calculate the evaluation index for indicating classification performance in the following way:
Wherein, AR is evaluation index, and TNR is the ratio for being determined as single compression video;TPR is
It is determined as the ratio of weight contracting video.
The verification and measurement ratio of video compress detection calculates expression in the following way:
Wherein, n is the number of test and the different video sample of training.
The present embodiment selects 20 training and test to obtain the average value of verification and measurement ratio, the detection of specific video compress detection
The average value of rate calculates in the following way:
Wherein, AR is evaluation index, n=20.
In order to more clearly embody a kind of visual classification feature extracting method and video code conversion weight provided by the present invention
The visual classification feature extracting method and video code conversion weight of the present invention are respectively adopted in the present embodiment for the advantage of compressed detected method
The co-occurrence matrix of compressed detected method block of pixels number corresponding with the PU classified types of I frames are used is as visual classification feature
Video code conversion weight contracting detection method is compared.Table 3 is the verification and measurement ratio of the weight contracting video of QCIF formats of the present invention, and table 3 is
The verification and measurement ratio of the weight contracting video of CIF formats of the present invention, table 5 are the corresponding block of pixels number of PU classified types using I frames
Detection of the co-occurrence matrix as the weight contracting video of the QCIF formats of the video code conversion weight contracting detection method of visual classification feature
Rate.
The verification and measurement ratio of the weight contracting video of 3. QCIF formats of the present invention of table
B1/B2 | 200k | 300k | 400k |
100k | 0.9667 | 0.9167 | 0.9125 |
200k | 0.9208 | 0.9750 | 0.9750 |
300k | 0.9208 | 0.9417 | 0.9500 |
The verification and measurement ratio of the weight contracting video of 4. CIF formats of the present invention of table
B1/B2 | 3.5M | 4M | 4.5M |
3M | 0.9813 | 0.9781 | 0.9875 |
3.5M | 0.9875 | 0.9750 | 0.9688 |
4M | 0.9813 | 0.9844 | 0.9813 |
Table 5. uses co-occurrence matrix the regarding as visual classification feature of the corresponding block of pixels number of PU classified types of I frames
The verification and measurement ratio of the weight contracting video of the QCIF formats of frequency transcoding weight contracting detection method
B1/B2 | 200k | 300k | 400k |
100k | 0.7750 | 0.8417 | 0.8667 |
200k | 0.8375 | 0.8709 | 0.8667 |
300k | 0.7957 | 0.8375 | 0.8917 |
As can be seen that visual classification feature extracting method using the present invention and the contracting of video code conversion weight from table 3 and table 4
The weight contracting verification and measurement ratio of detection method, QCIF and CIF formats has reached 90% or more, up to 98.75%, minimum
92.08%.
As can be seen from Table 5, in the prior art using the symbiosis of the corresponding block of pixels number of PU classified types using I frames
Video code conversion weight contracting detection method of the matrix as visual classification feature, the weight contracting video detection rate of QCIF formats exist
Between 77%-90%.Weight contracting video detection accuracy of the present invention is between 91%-97.5%, hence it is evident that is higher than the prior art.Together
When, using the visual classification feature extracting method and video code conversion weight contracting detection method of invention, PU classified types in embodiment
It is 25 kinds, uses the co-occurrence matrix of the corresponding block of pixels number of PU classified types of I frames as visual classification spy in the prior art
The PU classified types of the video code conversion weight contracting detection method of sign are 100 kinds.The PU classified types of the present invention are the prior art
1/4.The present invention is less than the prior art in PU classified types dimensions, is reducing calculation amount simultaneously, is improving weight contracting video
Verification and measurement ratio.A kind of visual classification feature extracting method of the present invention and video code conversion weight contracting detection method are more effective.
A kind of visual classification feature extracting method and video code conversion weight contracting detection method provided by the invention, for
H.264 weight compressed detected is carried out to the video code conversion weight contracting video of HEVC standard, the characteristic of division dimension of extraction is few, Neng Gouda
To higher verification and measurement ratio.
Each section of the present invention can be realized with hardware, software, firmware or combination thereof.In the above embodiment
In, software or firmware that multiple steps or method can in memory and by suitable instruction execution system be executed with storage come
It realizes.For example, if realized with hardware, in another embodiment, the known following technology in this field can be used
Any one of or their combination realize:With for data-signal realize logic function logic gates from
Logic circuit is dissipated, the application-specific integrated circuit with suitable combinational logic gate circuit, programmable gate array (PGA), scene can compile
Journey gate array (FPGA) etc..
The logic and/or step for indicating or describing in other ways herein in flow charts, for example, being considered
For realizing the order list of the executable instruction of logic function, any computer readable storage medium may be embodied in
In, for instruction execution system, device or equipment (system of such as computer based system including processor or other can be with
From instruction execution system, device or equipment instruction fetch and the system that executes instruction) use, or combine these instruction execution systems,
Device or equipment and use." computer readable storage medium " may include any medium for capableing of storage or transmission information.Machine
The example of device readable medium include electronic circuit, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disk,
CD-ROM, CD, hard disk, fiber medium, radio frequency (RF) link, etc..
As above it describes for one embodiment and/or the feature that shows can be in a manner of same or similar at one or more
It is used in a number of other embodiments, and/or the feature in other embodiments is combined or substitutes with the feature in other embodiments
It uses.
It should also be noted that, the exemplary embodiment referred in the present invention, is retouched based on a series of step or device
State certain methods or system.But the present invention is not limited to the sequence of above-mentioned steps, that is to say, that can be according in embodiment
The sequence referred to executes step, may also be distinct from that the sequence in embodiment or several steps are performed simultaneously.
In addition, method provided by the invention is equally readily applicable to compressed format standard transcoding before other HEVC standards
For the scene of HEVC formats, for example, be readily applicable to H.261, H.263, H.263+, in MPEG-1, MPEG-2 and MPEG-4
Any standard.
Explanation in conjunction with the present invention disclosed here and practice, the other embodiment of the present invention is for those skilled in the art
It all will be readily apparent and understand.Illustrate and embodiment is regarded only as being exemplary, true scope of the invention and purport are equal
It is defined in the claims.
Claims (9)
1. a kind of visual classification feature extracting method, which is characterized in that the method includes:
The predicting unit PU classified types of video frame are extracted using Visual Analyzer, and by the PU for the video frame extracted
Classified types are marked by basic unit of block of pixels;
Count the corresponding block of pixels number of each PU classified types of first P frame in each group of continuous pictures in video;
By the corresponding block of pixels number averaged of each PU classified types of first P frame in each group of continuous pictures, obtain
The characteristic of division of each PU classified types of first P frame in all groups of continuous pictures.
2. according to the method described in claim 1, it is characterized in that, the PU classified types of the video frame are with N × N block of pixels
Basic unit is marked, the integer multiple that wherein N is 4 or 4.
3. extracting method according to claim 1, which is characterized in that first P frame in each group of continuous pictures
The corresponding block of pixels number averaged of each PU classified types is realized by following formula:
Wherein Pi={ pi,0,pi,1..., pi,24(i=1,2 ..., M), M is the continuous pictures for including in video
Group number.
4. a kind of video code conversion weight contracting detection method, which is characterized in that the method includes:
It randomly selects the same number of single compression video and weight contracting video is sent into support vector machines as training sample;
According to the visual classification feature extracting method as described in any one of claim 1-3 to the single compression video and
Weight contracting video carries out visual classification feature extraction;
The support vector machines builds decision function according to the characteristic of division of extraction;
It randomly selects and is sent into the support vector machines as test sample for the single compression video and weight contracting video of test,
The video that the support vector machines exports discriminating test according to the decision function is that single compresses video or weight contracting video
Classification and Detection result.
5. according to the method described in claim 4, it is characterized in that, the method further includes:
The evaluation index for indicating classification performance is calculated in the following way:
Wherein, AR is evaluation index, and TNR is the ratio for being determined as single compression video;TPR is judgement
Attach most importance to and compresses the ratio of video.
6. according to the method described in claim 5, it is characterized in that, the method further includes:
The verification and measurement ratio for indicating video compress detection is calculated in the following way:
Wherein, n is the number of test and the different video sample of training.
7. according to the method described in claim 4, it is characterized in that, the weight contracting video be original video with the first bit rate
Carry out H.261, H.263, H.263+, H.264, any one of MPEG-1, MPEG-2 and MPEG-4 reference format compressed,
The video that HEVC compresses is carried out with the second bit rate to decoded video again after decoded.
8. according to the method described in claim 4, it is characterized in that, the singly compression video is original video with the second bit rate
Carry out the video that HEVC compresses.
9. a kind of computer readable storage medium, which is characterized in that be stored with computer journey in the computer readable storage medium
Sequence is performed the method and step realized as described in any one of claim 1-8 in the computer program.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810144370.0A CN108366295B (en) | 2018-02-12 | 2018-02-12 | Video classification feature extraction method, transcoding recompression detection method and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810144370.0A CN108366295B (en) | 2018-02-12 | 2018-02-12 | Video classification feature extraction method, transcoding recompression detection method and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108366295A true CN108366295A (en) | 2018-08-03 |
CN108366295B CN108366295B (en) | 2020-07-14 |
Family
ID=63006028
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810144370.0A Expired - Fee Related CN108366295B (en) | 2018-02-12 | 2018-02-12 | Video classification feature extraction method, transcoding recompression detection method and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108366295B (en) |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109587498A (en) * | 2018-12-25 | 2019-04-05 | 华南农业大学 | A kind of identification method by AVC transcoding HEVC video |
CN109982071A (en) * | 2019-03-16 | 2019-07-05 | 四川大学 | The bis- compression video detecting methods of HEVC based on time space complexity measurement and local prediction residual distribution |
CN111837140A (en) * | 2018-09-18 | 2020-10-27 | 谷歌有限责任公司 | Video coded field consistent convolution model |
CN113065533A (en) * | 2021-06-01 | 2021-07-02 | 北京达佳互联信息技术有限公司 | Feature extraction model generation method and device, electronic equipment and storage medium |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006279573A (en) * | 2005-03-29 | 2006-10-12 | Sanyo Electric Co Ltd | Encoder and encoding method, and decoder and decoding method |
CN101790097A (en) * | 2010-03-05 | 2010-07-28 | 天津大学 | Method for detecting multiple times of compression and coding of digital video |
CN102413327A (en) * | 2011-11-10 | 2012-04-11 | 中山大学深圳研究院 | Video tampering detection method based on compressed sensing semi fragile watermark |
US20120134415A1 (en) * | 2010-11-29 | 2012-05-31 | Mediatek Inc. | Method and Apparatus of Extended Motion Vector Predictor |
CN105072444A (en) * | 2015-07-31 | 2015-11-18 | 宁波大学 | Detection method for HEVC video secondary compression with different quantization parameters |
-
2018
- 2018-02-12 CN CN201810144370.0A patent/CN108366295B/en not_active Expired - Fee Related
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2006279573A (en) * | 2005-03-29 | 2006-10-12 | Sanyo Electric Co Ltd | Encoder and encoding method, and decoder and decoding method |
CN101790097A (en) * | 2010-03-05 | 2010-07-28 | 天津大学 | Method for detecting multiple times of compression and coding of digital video |
US20120134415A1 (en) * | 2010-11-29 | 2012-05-31 | Mediatek Inc. | Method and Apparatus of Extended Motion Vector Predictor |
CN102413327A (en) * | 2011-11-10 | 2012-04-11 | 中山大学深圳研究院 | Video tampering detection method based on compressed sensing semi fragile watermark |
CN105072444A (en) * | 2015-07-31 | 2015-11-18 | 宁波大学 | Detection method for HEVC video secondary compression with different quantization parameters |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111837140A (en) * | 2018-09-18 | 2020-10-27 | 谷歌有限责任公司 | Video coded field consistent convolution model |
CN109587498A (en) * | 2018-12-25 | 2019-04-05 | 华南农业大学 | A kind of identification method by AVC transcoding HEVC video |
CN109587498B (en) * | 2018-12-25 | 2022-07-08 | 华南农业大学 | Identification method for transcoding HEVC video by AVC |
CN109982071A (en) * | 2019-03-16 | 2019-07-05 | 四川大学 | The bis- compression video detecting methods of HEVC based on time space complexity measurement and local prediction residual distribution |
CN113065533A (en) * | 2021-06-01 | 2021-07-02 | 北京达佳互联信息技术有限公司 | Feature extraction model generation method and device, electronic equipment and storage medium |
CN113065533B (en) * | 2021-06-01 | 2021-11-02 | 北京达佳互联信息技术有限公司 | Feature extraction model generation method and device, electronic equipment and storage medium |
Also Published As
Publication number | Publication date |
---|---|
CN108366295B (en) | 2020-07-14 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
JP5039777B2 (en) | Quantization adjustment for DC shift artifacts | |
JP5124562B2 (en) | Quantization adjustment based on texture level | |
CN108632625B (en) | Video coding method, video decoding method and related equipment | |
Ravi et al. | Compression noise based video forgery detection | |
CN107657228B (en) | Video scene similarity analysis method and system, and video encoding and decoding method and system | |
CN108366295A (en) | Visual classification feature extracting method, transcoding weight contracting detection method and storage medium | |
US6473459B1 (en) | Scene change detector | |
US7184603B2 (en) | System and method for lossless compression of digital images | |
KR101695681B1 (en) | Context-based adaptive binary arithmetic coding (cabac) video stream compliance | |
US6058210A (en) | Using encoding cost data for segmentation of compressed image sequences | |
EP2782340B1 (en) | Motion analysis method based on video compression code stream, code stream conversion method and apparatus thereof | |
CN109104609B (en) | Shot boundary detection method fusing HEVC (high efficiency video coding) compression domain and pixel domain | |
Aghamaleki et al. | Inter-frame video forgery detection and localization using intrinsic effects of double compression on quantization errors of video coding | |
Stütz et al. | Non-blind structure-preserving substitution watermarking of H. 264/CAVLC inter-frames | |
Weng et al. | Optimal PPVO-based reversible data hiding | |
Xu et al. | Detection of double MPEG-2 compression based on distributions of DCT coefficients | |
KR101149522B1 (en) | Apparatus and method for detecting scene change | |
Bhardwaj et al. | A JPEG blocking artifact detector for image forensics | |
CN103067713B (en) | Method and system of bitmap joint photographic experts group (JPEG) compression detection | |
WO2016033725A1 (en) | Block segmentation mode processing method in video coding and relevant apparatus | |
Yao et al. | Double compression detection for H. 264 videos with adaptive GOP structure | |
US20080253670A1 (en) | Image Signal Re-Encoding Apparatus And Image Signal Re-Encoding Method | |
Wang et al. | Motion vector reversion-based steganalysis revisited | |
CN104410863B (en) | Image processor and image processing method | |
CN109862207B (en) | KVM video content change detection method based on compressed domain |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | 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 | ||
CF01 | Termination of patent right due to non-payment of annual fee | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20200714 Termination date: 20210212 |