CN104661023B - Image or method for video coding based on predistortion and training wave filter - Google Patents
Image or method for video coding based on predistortion and training wave filter Download PDFInfo
- Publication number
- CN104661023B CN104661023B CN201510058184.1A CN201510058184A CN104661023B CN 104661023 B CN104661023 B CN 104661023B CN 201510058184 A CN201510058184 A CN 201510058184A CN 104661023 B CN104661023 B CN 104661023B
- Authority
- CN
- China
- Prior art keywords
- image
- training
- wave filter
- predistortion
- video
- 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.)
- Expired - Fee Related
Links
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The present invention relates to image, video compression coding, code efficiency can either be improved to provide one kind, ringing effect is avoided again, so as to ensure the image or method for video coding based on predistortion and training wave filter of picture quality, image or method for video coding based on predistortion and training wave filter, by low pass filter, predistortion Fuzzy processing first is carried out to every frame video sequence before the coding；In decoding end, training sample database is formed using the largely image with various picture structures first；Then, off-line training is carried out to wave filter group with sorted image；After decoding end reconstructs blurred picture, picture structure is classified with the method that another sorting technique is combined using ADRC, optimal filter coefficients form optimal filter according to corresponding to classification results are found in a lookup table, distinguish deblurring by picture structure classification, be finally synthesizing to obtain the image after deblurring.Present invention is mainly applied to image, video compression coding.
Description
Technical field
The present invention relates to image, video compression coding, more particularly to based on predistortion and trains the image of wave filter or regards
Frequency coding method.
Background technology
With electronic equipment, the development of internet and the continuous improvement of people's demand, digital picture and video data are just
Produce and propagate at a terrific speed, high definition or ultra high-definition video have become development trend.For example, 3D, HDR (high dynamic
Scope), 4K or 8K videos etc., store or transmit these video datas and be required to efficient compression method to reduce data volume simultaneously
Ensure the quality of image.
Major video compression standard mainly have MPEG-2, MPEG-4, H.264/AVC, HEVC etc..These compression and coding standards
Using video sequence spatial coherence and temporal correlation, bulk redundancy information present in video sequence is removed, is only retained few
Measure irrelevant information to be transmitted, to reduce code check, save transmission bandwidth.And receiver utilizes these irrelevant informations, press
According to certain decoding algorithm, original video sequence content can be recovered on the premise of certain picture quality is ensured.Video compress
Estimation and dct transform in coding are the methods of important reduction signal redundancy.Estimation make use of the local of image
Directionality architectural characteristic, DCT make use of the grading structure characteristic that natural image has, i.e. low-pass characteristic, it is intended to remove picture in block
Redundancy between element.Dct transform can concentrate on energy on a small amount of low frequency coefficient, be easier to realize with reference to quantization and entropy code
Compression.These methods both contribute to the reduction of amount of coded data, in view of multi-medium data amount is increasing at present, are deposited in limited
Under the conditions of storage and bandwidth resources, further improving video compression coding efficiency has important realistic meaning.
2007, Lei.Zhichun [1] proposed a kind of video compression coding scheme, and this scheme is in coding side predistortion
Fuzzy processing video, and ambiguity function is transmitted in code stream, when decoding end reconstructs with this ambiguity function deconvolution deblurring.
The method can effectively reduce video data to be encoded amount and then reduce video stream code rate, but because using deconvolution in reconstruct
Deblurring can produce ringing effect.
Ringing effect is the inherent shortcoming of deconvolution deblurring method.In image processing field, researcher endeavours always
Deconvolution is replaced in finding a kind of new deblurring algorithm.1998, T.Kondo et al. [2] proposed one kind and is based on image
The least-mean-square filter of classification, there is preferable effect for image enhaucament.It is proposed to be classified according to picture structure again afterwards
Adaptive dynamic range coding (ADRC) algorithm to recover original image [3].2008, Ling Shao et al. [4] were proposed
ADRC recovers image with the algorithm that another image classification method is combined, and is directed to ADRC and mean absolute difference (MAD), ADRC
The algorithm being combined with standard deviation (STD), ADRC and dynamic range (DR) is tested, and effect is better than cascading filter.
2014, nightstool raised et al. [5] and proposes a kind of boundary effect restrainable algorithms, utilized the convolution pyramid filter with symmetry coefficient
Ripple device group model, by the respectively one group of filter coefficient of training, and this training is obtained of each type of region in certain particular image
Wave filter group be applied to solve other image respective types continuation region, the method calculating speed is fast, can effectively suppress each
The ringing effect of kind Frequency Domain Deconvolution algorithm.
In summary, studied proves that coding side predistortion Fuzzy Processing can reduce code check, training wave filter respectively
Method can realize image deblurring and avoid ringing effect (only limiting image procossing, be not directed to encode).At present it is not yet found that will
This two methods combines for image and video coding field.
Therefore, the present invention proposes a kind of image or Video Coding Scheme based on predistortion and training wave filter, that is, is compiling
Code end reduces code check using predistortion (blurring) processing, carries out deblurring using the good wave filter of off-line training in decoding end
Processing recovers the compression coding scheme of picture quality.
Bibliography
[1]L.Zhichun.Signal coding and decoding with pre-and post-processing.
Europe, 06006924.2 [P], 2007-10-03.
[2]T.Kondo and K.Kawaguchi,Adaptive dynamic range encoding method and
apparatus.U.S.patent 5444487,Aug.1998.
[3]T.Kondo,Y.Node,T.Fujiwara,and Y.Okumura,Picture conversion
apparatus.picture conversion method,learning apparatus and learning method,
U.S.patent 6323905,Nov.2001.
[4]Ling Shao,Hui Zhang,and Gerard de Haan.An Overview and Performance
Evaluation of Classification-Base Least Squares Trained Filters[C].//
Transaction on Image Processing,2008,17(10):1772-1782.
[5] nightstool is raised, Liu Xuehui, quick restrainable algorithms [J] computer aided manufacturings of Wu's grace China image deconvolution boundary effects
Help design and graphics journal, 2014,26 (7)：1051-1066.
The content of the invention
For overcome the deficiencies in the prior art, there is provided one kind can either improve code efficiency, avoid ringing effect again, so as to protect
Demonstrate,prove the image or method for video coding based on predistortion and training wave filter of picture quality.Therefore, the technology that the present invention takes
Scheme is image or method for video coding based on predistortion and training wave filter, first before the coding by low pass filter
Predistortion Fuzzy processing is carried out to every frame video sequence；
In decoding end, training sample database is formed using the largely image with various picture structures first, according to image knot
The difference of structure is classified to different zones, specific that image is carried out with the method that another sorting technique is combined using ADRC
Classification；
Then, off-line training is carried out to wave filter group with sorted image, using the training algorithm of least mean-square error,
For each type of region, respectively one group of optimal filter coefficient of training, formation look-up table (LUT), deposit decoding end reconstruct mould
Block；
After decoding end reconstructs blurred picture, the method being combined using ADRC with another sorting technique is to image
Structure is classified, and optimal filter coefficients form optimal filter according to corresponding to classification results are found in a lookup table, press
Picture structure classification distinguishes deblurring, is finally synthesizing to obtain the image after deblurring.
Picture structure specifically includes smooth region, texture region, fringe region.
Another sorting technique is specially in local entropy, mean absolute difference (MAD), standard deviation (STD), dynamic range (DR)
One kind.
In decoded reconstructed image, if running into the picture structure not having in training sample database, i.e., no pair in look-up table
The filter coefficient of this structure is answered, then replaces training wave filter to recover image or video using Deconvolution Method, i.e., by warp
Product method is as a kind of alternative scheme outside training wave filter deblurring algorithm, for recovering original image or video.
Compared with the prior art, technical characterstic of the invention and effect：
The present invention proposes image or Video Coding Scheme based on predistortion and training wave filter, the predistortion mould of coding side
Paste processing can reduce code check, improve code efficiency.The training wave filter deblurring algorithm at decoding and reconstituting end is according to picture structure
Difference carry out deblurring from different wave filter and recover original image, ringing effect can be avoided, improve decoding image or
Video quality.
Brief description of the drawings
Fig. 1 image or Video Coding Scheme proposed by the present invention based on predistortion and training wave filter
Embodiment
Multi-medium data amount is increasing, while consumer requires more and more higher to multimedia quality.Therefore need to study
Code efficiency can further be improved and ensure the algorithm of certain coding quality.
Data to be encoded amount, raising code efficiency can be reduced by carrying out predistortion Fuzzy processing in coding side, but to decoding
The deblurring function of end reconstructed module proposes challenge.If transmission ambiguity function can increase code stream expense, and traditional warp
Product image deblurring algorithm can cause ringing effect.Therefore, exploitation has suitable for predistortion-deblurring scheme of Video coding
Significance.
The present invention proposes a kind of image or Video Coding Scheme based on predistortion and training wave filter.This scheme is encoding
End carries out predistortion (blurring) processing to image or video, and without transmitting coding side predistortion (blurring) letter in code stream
Number, deblurring is carried out from the good different wave filters of off-line training according to different images structure in decoding end.It can either carry
High coding efficiency, ringing effect is avoided again, ensure that picture quality.
In existing encoding scheme, generally, LPF is carried out before the coding to remove high-frequency noise, it is this low
The mode of pass filter does not have a significant effect to picture quality.Further to reduce data to be encoded amount, the present invention is in coding
It is preceding that predistortion Fuzzy processing first is carried out to every frame video sequence, the low pass filter that size is 5 × 5 can be passed to and realized,
Then coding is compressed again.The fuzzy high fdrequency component for causing image is greatly decreased, and compression efficiency is higher, but Fuzzy processing makes
Image fault using deblurring algorithm when decoding end reconstructs, it is necessary to recover picture quality.
The algorithm for training wave filter deblurring is used for decoding end reconstructed module by the present invention.It is various using largely having first
The image composition training sample database of picture structure (smooth region, texture region, fringe region etc.), according to the difference of picture structure
Different zones are classified, usually, the method for adaptive dynamic range coding (ADRC) can be used to tie the difference of image
Structure is classified, but the method does not distinguish the difference of image high-contrast and low contrast structure and do not account for object
The problem of detail section compression noise is identical, therefore image can be entered with the method that another sorting technique is combined using ADRC
Row classification.For example, ADRC and local entropy, ADRC and mean absolute difference (MAD), ADRC and standard deviation (STD), ADRC and dynamic model
Enclose method that (DR) is combined etc..
Then, off-line training is carried out to wave filter group with sorted image, the training of least mean-square error can be used
Algorithm, one group of optimal filter coefficient is respectively trained for each type of region, forms look-up table (LUT), is stored in decoding end
Reconstructed module.
After decoding end reconstructs blurred picture, using ADRC with another sorting technique is combined (such as local entropy)
Method is classified to picture structure, and optimal filter coefficients composition is optimal according to corresponding to classification results are found in a lookup table
Wave filter, distinguish deblurring by picture structure classification, be finally synthesizing to obtain the image after deblurring.
In addition, in decoded reconstructed image, (do not have if running into the picture structure not having in training sample database in look-up table
Have the filter coefficient for corresponding to this structure), Deconvolution Method can be used to replace training wave filter to recover image or video, i.e.,
Deconvolution Method can as training wave filter deblurring algorithm outside a kind of alternative scheme, for recover original image or
Video.
The present invention proposes a kind of image or Video Coding Scheme based on predistortion and training wave filter.Wherein, filtering
In terms of device training, using adaptive dynamic range encode (ADRC) and local entropy (or ADRC and mean absolute difference (MAD),
ADRC and standard deviation (STD), ADRC and dynamic range (DR)) method that is combined classified for the different structure of image,
The training algorithm of least mean-square error is used afterwards, and for each type of region, respectively one group of filter coefficient of training, formation are looked into
Table (LUT) is looked for, is stored in decoding end reconstructed module.In the implementation process of whole scheme, first, coding side to input picture or
Person's video sequence carries out pre-distortion to reduce code check；Then according to existing encoding scheme, to the image after predistortion or regard
Frequency carries out encoding and decoding；Finally, in decoded reconstructed image, using the wave filter of the completion of off-line training, according to picture structure
It is different that optimal filter factor is searched in look-up table (LUT), complete the reconstruction of image or video.
Claims (4)
1. a kind of image or method for video coding based on predistortion and training wave filter, it is characterized in that, comprise the following steps：
In coding side：By low pass filter, predistortion Fuzzy processing first is carried out to every frame video sequence before the coding；
In decoding end, training sample database is formed using the largely image with various picture structures first, according to picture structure
Difference is classified to different zones, it is specific using the adaptive dynamic range classified according to picture structure encode ADRC with
The method that another sorting technique is combined is classified to image；
Then, off-line training is carried out to wave filter group with sorted image, using the training algorithm of least mean-square error, for
One group of optimal filter coefficient is respectively trained in each type of region, forms look-up table (LUT), is stored in decoding end reconstructed module；
After decoding end reconstructs blurred picture, the method being combined using ADRC with another sorting technique is to picture structure
Classified, optimal filter coefficients form optimal filter according to corresponding to classification results are found in a lookup table, by image
Structured sort distinguishes deblurring, is finally synthesizing to obtain the image after deblurring.
2. image or method for video coding according to claim 1 based on predistortion and training wave filter, it is characterized in that,
Picture structure specifically includes smooth region, texture region, fringe region.
3. image or method for video coding according to claim 1 based on predistortion and training wave filter, it is characterized in that,
Another sorting technique is specially one kind in local entropy, mean absolute difference (MAD), standard deviation (STD), dynamic range (DR).
4. image or method for video coding according to claim 1 based on predistortion and training wave filter, it is characterized in that,
In decoded reconstructed image, if running into the picture structure not having in training sample database, i.e., without this corresponding structure in look-up table
Filter coefficient, then replace training wave filter deblurring algorithm to recover image or video using Deconvolution Method, will be anti-
Convolution method is as a kind of alternative scheme outside training wave filter deblurring algorithm, for recovering original image or video.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510058184.1A CN104661023B (en) | 2015-02-04 | 2015-02-04 | Image or method for video coding based on predistortion and training wave filter |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510058184.1A CN104661023B (en) | 2015-02-04 | 2015-02-04 | Image or method for video coding based on predistortion and training wave filter |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104661023A CN104661023A (en) | 2015-05-27 |
| CN104661023B true CN104661023B (en) | 2018-03-09 |
Family
ID=53251651
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510058184.1A Expired - Fee Related CN104661023B (en) | 2015-02-04 | 2015-02-04 | Image or method for video coding based on predistortion and training wave filter |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104661023B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020097888A1 (en) * | 2018-11-15 | 2020-05-22 | 深圳市欢太科技有限公司 | Video processing method and apparatus, electronic device, and computer-readable storage medium |
| CN110149554B (en) * | 2019-05-31 | 2021-06-15 | Oppo广东移动通信有限公司 | Video image processing method and device, electronic equipment and storage medium |
| CN110598584A (en) * | 2019-08-26 | 2019-12-20 | 天津大学 | Convolutional neural network face recognition algorithm based on wavelet transform and DCT |
| CN113723472B (en) * | 2021-08-09 | 2023-11-24 | 北京大学 | Image classification method based on dynamic filtering constant-variation convolutional network model |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP0601878A1 (en) * | 1992-12-10 | 1994-06-15 | Sony Corporation | Adaptive dynamic range encoding method and apparatus |
| US6323905B1 (en) * | 1997-12-25 | 2001-11-27 | Sony Corporation | Picture conversion apparatus picture conversion method learning apparatus and learning method |
| CN101325655A (en) * | 2007-06-15 | 2008-12-17 | 索尼株式会社 | Image signal processing apparatus and method, image display and output apparatus |
| CN102413330A (en) * | 2007-06-12 | 2012-04-11 | 浙江大学 | Texture-adaptive video coding/decoding system |
| CN103475876A (en) * | 2013-08-27 | 2013-12-25 | 北京工业大学 | Learning-based low-bit-rate compression image super-resolution reconstruction method |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP1840875A1 (en) * | 2006-03-31 | 2007-10-03 | Sony Deutschland Gmbh | Signal coding and decoding with pre- and post-processing |
| JP4835949B2 (en) * | 2007-12-21 | 2011-12-14 | ソニー株式会社 | Image processing apparatus and method, learning apparatus and method, program, and recording medium |
-
2015
- 2015-02-04 CN CN201510058184.1A patent/CN104661023B/en not_active Expired - Fee Related
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP0601878A1 (en) * | 1992-12-10 | 1994-06-15 | Sony Corporation | Adaptive dynamic range encoding method and apparatus |
| US6323905B1 (en) * | 1997-12-25 | 2001-11-27 | Sony Corporation | Picture conversion apparatus picture conversion method learning apparatus and learning method |
| CN102413330A (en) * | 2007-06-12 | 2012-04-11 | 浙江大学 | Texture-adaptive video coding/decoding system |
| CN101325655A (en) * | 2007-06-15 | 2008-12-17 | 索尼株式会社 | Image signal processing apparatus and method, image display and output apparatus |
| CN103475876A (en) * | 2013-08-27 | 2013-12-25 | 北京工业大学 | Learning-based low-bit-rate compression image super-resolution reconstruction method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104661023A (en) | 2015-05-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107197260B (en) | Video coding post-filter method based on convolutional neural networks | |
| Wang et al. | Attention-based dual-scale CNN in-loop filter for versatile video coding | |
| CN104661023B (en) | Image or method for video coding based on predistortion and training wave filter | |
| CN101668196B (en) | Low code rate image compression method based on down sampling and interpolation | |
| CN111885280B (en) | Hybrid convolutional neural network video coding loop filtering method | |
| CN113055674B (en) | Compressed video quality enhancement method based on two-stage multi-frame cooperation | |
| CN105898300A (en) | Improved transformation coefficient sign data hiding method based on recovered transformation coefficient | |
| CN113592746B (en) | Method for enhancing quality of compressed video by fusing space-time information from coarse to fine | |
| Wang et al. | An integrated CNN-based post processing filter for intra frame in versatile video coding | |
| Ma et al. | Rate-perception optimized preprocessing for video coding | |
| Liu et al. | Convolutional neural network-based residue super-resolution for video coding | |
| CN115442613A (en) | Interframe information-based noise removal method using GAN | |
| CN104581173A (en) | Soft decoding verification model platform | |
| Shen et al. | Dec-adapter: Exploring efficient decoder-side adapter for bridging screen content and natural image compression | |
| CN105141967A (en) | Fast adaptive loop filter algorithm based on just noticeable distortion (JND) model | |
| Qi et al. | CNN-based post-processing filter for video compression with multi-scale feature representation | |
| Del Corso et al. | MNR: A novel approach to correct MPEG temporal distortions | |
| Khan et al. | Performance analysis of H. 264 video decoder: Algorithm and applications | |
| JP2015080115A (en) | Image encoding device, image encoding method, and program | |
| Man et al. | Deep learning-assisted video compression framework | |
| Palaparthi et al. | A simple deblocking method for reduction of blocking artifacts | |
| Zhang et al. | Encoder-Decoder Joint Enhancement for Video Chat | |
| CN103209323A (en) | Image compression method based on vectors among k<th>-order zero trees | |
| Wei et al. | New method for reducing boundary artifacts in block-based wavelet image compression | |
| Zhao et al. | Adaptive Amplitude Modulation and Recovery Method for Video Coding |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20180309 Termination date: 20210204 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |