CN104994397B - Adaptive video preprocess method based on time-domain information - Google Patents
Adaptive video preprocess method based on time-domain information Download PDFInfo
- Publication number
- CN104994397B CN104994397B CN201510397733.8A CN201510397733A CN104994397B CN 104994397 B CN104994397 B CN 104994397B CN 201510397733 A CN201510397733 A CN 201510397733A CN 104994397 B CN104994397 B CN 104994397B
- Authority
- CN
- China
- Prior art keywords
- pixel
- value
- bilateral filtering
- time
- domain information
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- 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
- H04N19/134—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or criterion affecting or controlling the adaptive coding
- H04N19/154—Measured or subjectively estimated visual quality after decoding, e.g. measurement of distortion
-
- 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
- H04N19/85—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using pre-processing or post-processing specially adapted for video compression
Abstract
The present invention discloses a kind of adaptive video preprocess method based on time-domain information, and steps are as follows:Step S1, a frame video image data are read in;Step S2, each weighting coefficient of the pixel based on time-domain information is obtained;Step S3, the bilateral filtering value of each pixel is calculated;Step S4, according to weighting coefficient and bilateral filtering value, the value for the pixel that allows treated is taken from the weighted sum of original pixel and the pixel bilateral filtering value, obtains video pre-filtering value.The present invention is based on the bilateral filtering values and weighting coefficient of each pixel to handle original image, the value obtained is the weighted sum of bilateral filtering value and original pixel value, it is accurate and opposing stationary region keeps its time-domain information constant in estimation to realize, estimation acutely and residual error estimate relatively inaccurate region by bilateral filtering realize noise reduction boundary preservation effect, the relativity of time domain between frame before and after video can farthest be retained, and obtain apparent code check simultaneously and decline and the promotion of subjective quality.
Description
Technical field
The present invention relates to video pre-filtering field, the adaptive video pretreatment side based on time-domain information is especially related to
Method.
Technical background
With the development of the development of information technology, especially HD video and mobile Internet, the quantity of multimedia video
And quickly expansion is presented in demand.Standard of the development compared with previous generation of video encoding standard HEVC of new generation has reduced
50% code check, the next-generation new video encoding standard of development are simultaneously taken to and still need longer development time in actual application,
And by combining the technology of video pre-filtering, the efficiency of Video coding can be obviously improved and promoting its subjective quality.
Video pre-filtering technology is a series of processing operations carried out to it before video is encoded, and main purpose is
In order to reduce code check and promote quality.Develop by years of researches, video pre-filtering technology is mainly filtered, inserted including traditional
The methods of value and de interlacing.The preprocess method that coding and JND based on area-of-interest are combined can be promoted effectively
The quality and efficiency of Video coding.
In common video pre-filtering technology, the processing of spatial information (si) is easy to destroy the time domain phase between front and back frame
Guan Xing, to which the effect of video pre-filtering can be reduced.
Invention content
The technical problem to be solved by the invention is to provide a kind of adaptive video pretreatment side based on time-domain information
Method farthest to retain the relativity of time domain between front and back frame, and reduces code check, promotes subjective quality.
In order to solve the above technical problems, the present invention adopts the following technical scheme that:A kind of adaptively regarding based on time-domain information
Frequency preprocess method, its step are as follows:
Step S1, a frame video image data are read in;
Step S2, each weighting coefficient of the pixel based on time-domain information is obtained, it is residual based on operation estimation to calculate each pixel of acquisition
The weighting coefficient of difference, in conjunction with the information of estimation, weighting coefficient wt(p) calculation formula is as follows:
(1)
In formula(1)In, p is current pixel position, and Ω is the pixel region of the predefined size centered on p, and
Residuen(p) it is the residual error of the estimation of pixel in the regions Ω, Thresh1It is the parameter for promoting function smooth;
Step S3, the bilateral filtering value of each pixel is calculated;
Step S4, according to weighting coefficient and bilateral filtering value, the value of the pixel that allows treated be taken from original pixel and
The weighted sum of the pixel bilateral filtering value obtains video pre-filtering value.
Further, when carrying out step S1, to the first frame data without processing.
Further, in step s3, bilateral filtering value If(p) calculation formula is as follows:
(2)
In formula(2)In, ΩfFor the window size of bilateral filtering, In(p) it is original pixel value, and wd(p, q,σd) and wr
(p, q,σr) it is spatial domain core and codomain core respectively, it is calculated and is obtained by following formula respectively:
(3)
(4).
Further, in step s 4, the calculation formula of the pixel value after image procossing is as follows:
(5)
In formula(5)In, In(p) it is original pixel value, If(p) be the pixel bilateral filtering value, Thresh2For threshold value, when
When the residual sum of the estimation of pixel in the region is less than threshold value, then it is constant to retain its pixel value.
By using above-mentioned technical proposal, the present invention has the following technical effects:The present invention is believed by calculating based on time domain
The weighting coefficient of breath is based particularly on the weighting coefficient of estimation residual error, to which quantization meter is realized in the influence of time-domain information
It calculates, then original image is handled based on bilateral filtering value and weighting coefficient, the value after the image procossing obtained is bilateral filtering value
With the weighted sum of original pixel value, it is accurate and opposing stationary region keeps its time-domain information constant in estimation to realize,
Estimation acutely and residual error estimate relatively inaccurate region by bilateral filtering realize noise reduction boundary preservation effect,
The relativity of time domain between frame before and after video can farthest be retained, and obtain decline and the subjectivity of apparent code check simultaneously
The promotion of quality.
Description of the drawings
Fig. 1 is the frame diagram of the adaptive video preprocess method the present invention is based on time-domain information.
Fig. 2 is that the present invention is based on cycle tests under each resolution ratio of the adaptive video preprocess method of time-domain information
The change curve of code check income and average yield.
Fig. 3 be the present invention is based on the adaptive video preprocess method of time-domain information in 174kbps direct coding sequence
The schematic diagram of coded sequence frame after frame and pretreatment.
Fig. 4 is that the present invention is based on the adaptive video preprocess methods of time-domain information at code check 646kbps
PartyScene two regional correlation schematic diagrames of the 37th frame sequence.
Specific implementation mode
It should be noted that in the absence of conflict, the features in the embodiments and the embodiments of the present application can phase
It mutually combines, invention is further described in detail in the following with reference to the drawings and specific embodiments.
As shown in Figure 1, the present invention provides a kind of adaptive video preprocess method based on time-domain information, key step
It is as follows:
Step S1, a frame video image data are read in;
Step S2, each weighting coefficient of the pixel based on time-domain information is obtained;
Step S3, the bilateral filtering value of each pixel is calculated;
Step S4, video pre-filtering value is obtained according to weighting coefficient and bilateral filtering value.
The specific implementation process of each step is described in detail below.
Step S1
To ensure the accuracy of time-domain information, the first frame data can be given tacit consent to without processing.
Step S2
When performing the encoding operation, if pervious motion residuals can obtain the effect of clearing after carrying out change quantization,
And quantization transform may not be carried out after the filtering operation to 0, to weaken the effect for reducing code check.
In the motion estimation process of Video coding, when the residual error of estimation is very small and belong to stagnant zone
When, it was demonstrated that the accuracy of estimation is relatively high, residual error then can very little, after carrying out final change quantization, code rate information meeting
Seldom, at the same mass loss can very little, thus carry out it is pretreated during, to this kind of pixel or pixel region into
Row retains or carries out slight filtering operation;And the pixel for other complicated movements or region, intensity can be carried out
The filtering operation of bigger removes visual redundancy information, retains or strengthens key area, and then can reach reduction code check and be promoted
The effect of quality.
Due to the boundary retention characteristic of bilateral filtering, which is being used to pixel to be processed or region
Formula can not only remove redundancy, make estimation more accurate, on the other hand, the boundary retention characteristic of the processing mode,
It can maintain the important feature of image.
In order to allow original image after carrying out pretreatment operation, at the same it is relatively accurate to stagnant zone and estimation
Region retains more raw informations, while removing redundancy to nonstatic region and motion estimation error region relatively bigger than normal
Information still retains key feature information, and the present invention uses a weighting coefficient, and the value for the pixel that allows treated is taken from original
The weighted sum of pixel and the pixel bilateral filtering value, in conjunction with the information of estimation, weighting coefficient is as follows:
(1)
In formula(1)In, p is current pixel position, and Ω is the 3x3 pixel regions centered on p, and Residuen(p)
For the residual error of the estimation of pixel in the regions Ω, Thresh1It is the parameter in order to make the function more smooth, value model
Enclosing can be 400 ~ 600, preferably 500.As can be seen that when the pixel region is static or estimation is more accurate, it should
The motion residuals value quadratic sum meeting smaller in region, and then weighting coefficient w can be causedt(p) bigger, conversely, working as estimation residual error
When being worth bigger than normal, then weighting coefficient wt(p) will become smaller.
Step S3
When carrying out this step, If(p) bilateral filtering is defined as follows:
(2)
In formula(2)In, ΩfIt can be taken both horizontally and vertically in the specific implementation for the window size of bilateral filtering
± 3 location of pixels sizes, the i.e. size windows of 7x7, and wd(p, q,σd) and wr(p, q,σr) it is spatial domain core and codomain respectively
Core:
(3)
(4).
Step S4
It obtains and the relevant weighting coefficient w of time domaint(p) after, the pixel value after image procossing becomes:
(5)
In formula(5)In, In(p) it is original pixel value, If(p) be the pixel carry out bilateral filtering after value.It is residual when moving
Poor smaller and opposing stationary region, pretreated pixel value more tends to original pixel value, conversely, pretreated pixel value
More approach the pixel value of bilateral filtering.Pass through given threshold Thresh2, threshold value Thresh2Value range can be 40 ~ 60,
Preferably 50.When the residual sum in the region is less than threshold value, assert that the location of pixels estimation is accurate, it should retain its pixel
Be worth it is constant, meanwhile, also reduce computational complexity.
The experiment effect of the preprocess method of the present invention is as follows:Relative to H.264 equal encoders, HEVC estimate in movement in the past
More accurate and efficient prediction technique is taken on meter, thus in this experiment, when we calculate residual error, uses institute above
The method for estimating of the HEVC of description.Original video sequence and the pretreated video sequence of progress are compiled in HM 12.0 respectively
It is encoded on code device, compares its code check and subjective quality.Wherein in this experiment, parameter selection is respectively Thresh1=
500, Thresh2=50, σd=3, σr=15.Cycle tests has 416x240p, 832x480p, 1280x720p respectively,
1920x1080p.Between the QP of each cycle tests chooses from 31 to 40.
By using the preprocess method of the present invention, the average 6% code check decline lost without subjective quality can be obtained
(As shown in Figure 2), and the promotion of subjective quality can also be obtained in some video sequences, as shown in Figure 3 and Figure 4, wherein Fig. 3
In upper figure be image after direct coding, figure below is the image encoded after pretreatment;Lower section six in Fig. 4 is opened in small figure, left
The a groups of side are the image encoded after pre-processing, and the image that intermediate b groups are direct coding, and the c groups on right side are global bilateral filter
The image encoded after wave.
It although an embodiment of the present invention has been shown and described, for the ordinary skill in the art, can be with
Understanding without departing from the principles and spirit of the present invention can carry out these embodiments a variety of variations, modification, replace
And modification, the scope of the present invention are limited by appended claims and its equivalency range.
Claims (4)
1. a kind of adaptive video preprocess method based on time-domain information, which is characterized in that its step are as follows:
Step S1, a frame video image data are read in;
Step S2, each weighting coefficient of the pixel based on time-domain information is obtained, calculates and obtains each pixel based on operation estimation residual error
Weighting coefficient, in conjunction with the information of estimation, weighting coefficient wt(p) calculation formula is as follows:
(1)
In formula(1)In, p is current pixel position, and Ω is the pixel region of the predefined size centered on p, and Residuen
(p) it is the residual error of the estimation of pixel in the regions Ω, Thresh1It is the parameter for promoting function smooth;
Step S3, the bilateral filtering value of each pixel is calculated;
Step S4, according to weighting coefficient and bilateral filtering value, the value for the pixel that allows treated is taken from original pixel and the picture
The weighted sum of plain bilateral filtering value obtains video pre-filtering value.
2. the adaptive video preprocess method based on time-domain information as described in claim 1, which is characterized in that carry out step
When S1, to the first frame data without processing.
3. the adaptive video preprocess method based on time-domain information as described in claim 1, which is characterized in that in step S3
In, bilateral filtering value If(p) calculation formula is as follows:
(2)
In formula(2)In, ΩfFor the window size of bilateral filtering, In(p) it is original pixel value, and wd(p, q,σd) and wr(p, q,
σr) it is spatial domain core and codomain core respectively, it is calculated and is obtained by following formula respectively:
(3)
(4).
4. the adaptive video preprocess method based on time-domain information as claimed in claim 3, which is characterized in that in step S4
In, the calculation formula of the pixel value after image procossing is as follows:
(5)
In formula(5)In, In(p) it is original pixel value, If(p) be the pixel bilateral filtering value, Thresh2For threshold value, when the area
When the residual sum of the estimation of pixel in domain is less than threshold value, then it is constant to retain its pixel value.
Priority Applications (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510397733.8A CN104994397B (en) | 2015-07-08 | 2015-07-08 | Adaptive video preprocess method based on time-domain information |
PCT/CN2015/089105 WO2017004883A1 (en) | 2015-07-08 | 2015-09-08 | Time-domain information-based adaptive video pre-processing method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510397733.8A CN104994397B (en) | 2015-07-08 | 2015-07-08 | Adaptive video preprocess method based on time-domain information |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104994397A CN104994397A (en) | 2015-10-21 |
CN104994397B true CN104994397B (en) | 2018-11-09 |
Family
ID=54306135
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510397733.8A Active CN104994397B (en) | 2015-07-08 | 2015-07-08 | Adaptive video preprocess method based on time-domain information |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN104994397B (en) |
WO (1) | WO2017004883A1 (en) |
Families Citing this family (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105959663B (en) * | 2016-05-24 | 2018-09-21 | 厦门美图之家科技有限公司 | The successional optimized treatment method of video interframe signal, system and camera terminal |
US10931974B2 (en) | 2016-10-14 | 2021-02-23 | Mediatek Inc. | Method and apparatus of smoothing filter for ringing artefact removal |
CN112639867A (en) * | 2020-05-07 | 2021-04-09 | 深圳市大疆创新科技有限公司 | Image processing method and device |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1929552A (en) * | 2006-09-30 | 2007-03-14 | 四川长虹电器股份有限公司 | Spatial domain pixel data processing method |
CN102281439A (en) * | 2011-06-16 | 2011-12-14 | 杭州米加科技有限公司 | Streaming media video image preprocessing method |
CN103108109A (en) * | 2013-01-31 | 2013-05-15 | 深圳英飞拓科技股份有限公司 | Digital video noise reduction system and method |
CN103632356A (en) * | 2012-08-29 | 2014-03-12 | 华为技术有限公司 | Method and device for enhancing image spatial resolution |
US9300846B2 (en) * | 2012-09-10 | 2016-03-29 | Apple Inc. | Signal shaping for improved mobile video communication |
Family Cites Families (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
EP1100260A1 (en) * | 1999-11-12 | 2001-05-16 | STMicroelectronics S.r.l. | Spatio-temporal filtering method for noise reduction during pre-processing of picture sequences in video encoders |
US7394856B2 (en) * | 2003-09-19 | 2008-07-01 | Seiko Epson Corporation | Adaptive video prefilter |
US7657113B2 (en) * | 2005-12-21 | 2010-02-02 | Hong Kong Applied Science And Technology Research Institute Co., Ltd. | Auto-regressive method and filter for denoising images and videos |
CN101448077B (en) * | 2008-12-26 | 2010-06-23 | 四川虹微技术有限公司 | Self-adapting video image 3D denoise method |
CN103024248B (en) * | 2013-01-05 | 2016-01-06 | 上海富瀚微电子股份有限公司 | The video image noise reducing method of Motion Adaptive and device thereof |
CN103533214B (en) * | 2013-10-01 | 2017-03-22 | 中国人民解放军国防科学技术大学 | Video real-time denoising method based on kalman filtering and bilateral filtering |
-
2015
- 2015-07-08 CN CN201510397733.8A patent/CN104994397B/en active Active
- 2015-09-08 WO PCT/CN2015/089105 patent/WO2017004883A1/en active Application Filing
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN1929552A (en) * | 2006-09-30 | 2007-03-14 | 四川长虹电器股份有限公司 | Spatial domain pixel data processing method |
CN102281439A (en) * | 2011-06-16 | 2011-12-14 | 杭州米加科技有限公司 | Streaming media video image preprocessing method |
CN103632356A (en) * | 2012-08-29 | 2014-03-12 | 华为技术有限公司 | Method and device for enhancing image spatial resolution |
US9300846B2 (en) * | 2012-09-10 | 2016-03-29 | Apple Inc. | Signal shaping for improved mobile video communication |
CN103108109A (en) * | 2013-01-31 | 2013-05-15 | 深圳英飞拓科技股份有限公司 | Digital video noise reduction system and method |
Non-Patent Citations (2)
Title |
---|
基于像素层标识点得双边滤波快速实现;郑丽萍等;《计算机科学》;20130731;第40卷(第7期);第262-265页 * |
基于运动补偿的视频预处理算法;谢正光等;《计算机工程》;20091031;第35卷(第19期);第222-223页 * |
Also Published As
Publication number | Publication date |
---|---|
WO2017004883A1 (en) | 2017-01-12 |
CN104994397A (en) | 2015-10-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN104244015B (en) | For removing the method and system of the coding artifact in video data | |
Wu et al. | Learned block-based hybrid image compression | |
JP2006507775A (en) | Method and apparatus for measuring the quality of a compressed video sequence without criteria | |
CA2856634A1 (en) | Texture masking for video quality measurement | |
EP0791270A1 (en) | Spatially adaptive filtering for video encoding | |
Soh et al. | Reduction of video compression artifacts based on deep temporal networks | |
CN107820085B (en) | Method for improving video compression coding efficiency based on deep learning | |
WO2017004889A1 (en) | Jnd factor-based super-pixel gaussian filter pre-processing method | |
JP2015095764A (en) | Encoder and encoding method | |
CN104994397B (en) | Adaptive video preprocess method based on time-domain information | |
Meng et al. | Mganet: A robust model for quality enhancement of compressed video | |
CN108028937A (en) | video motion compensation device and method | |
Huang et al. | Neural compression-based feature learning for video restoration | |
Dai et al. | Film grain noise removal and synthesis in video coding | |
Meng et al. | Enhancing quality for VVC compressed videos by jointly exploiting spatial details and temporal structure | |
CN108141602A (en) | video motion compensation device and method | |
CN1650635A (en) | Method of processing digital images for low-rate applications | |
CN113055674B (en) | Compressed video quality enhancement method based on two-stage multi-frame cooperation | |
CN110351453A (en) | A kind of computer video data processing method | |
Hu et al. | Deep inter prediction with error-corrected auto-regressive network for video coding | |
Wu et al. | Perceptually weighted mean squared error based rate-distortion optimization for HEVC | |
CN114173131A (en) | Video compression method and system based on inter-frame correlation | |
Chen et al. | AV1 video coding using texture analysis with convolutional neural networks | |
CN105872538B (en) | Time domain filtering method and time domain filtering device | |
Del Corso et al. | MNR: A novel approach to correct MPEG temporal distortions |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |