CN103607591B - Video image compression method combining super-resolution reconstruction - Google Patents
Video image compression method combining super-resolution reconstruction Download PDFInfo
- Publication number
- CN103607591B CN103607591B CN201310518474.0A CN201310518474A CN103607591B CN 103607591 B CN103607591 B CN 103607591B CN 201310518474 A CN201310518474 A CN 201310518474A CN 103607591 B CN103607591 B CN 103607591B
- Authority
- CN
- China
- Prior art keywords
- image
- video image
- super
- video
- resolution
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention discloses a video image compression method combining super-resolution reconstruction. The invention combines traditional JPEG, JPEG2000, H.264 and HEVC-code standard algorithms with super-resolution image reconstruction and designs a video image compression method combining the super-resolution reconstruction on the basis. Downsampling is carried out on an input video image, wherein a downsampling method adopts a Bicubic algorithm and a downsampling multiple is 2. The number of dot arrays of a downsampling image is only 1/4 of that of an original image. The encoding rate of the downsampling image is far lower than that of an original input image so that the encoding rate is reduced. At the same time, on the basis that robustness differences of a residual image and a general image are analyzed, a negative-feedback step is introduced in the design so that part of high-frequency detail information lost in a super-resolution image reconstruction step is remedied and reconstructed video or image quality is improved. Compared with the JPEG and the H.264 standard algorithms, the compression method reduces the encoding rate greatly under a situation that image quality is the same.
Description
Technical field
The present invention relates to the image lossy compression in field of picture communication, especially relate to image super-resolution
A kind of new coded method being combined together with video and compression of images.
Background technology
Compression of images is the basis that digital multimedia calculates with communication technology, is wherein directed to the lossy compression method of video and image
Method is the emphasis studied for many years.Developing rapidly with Digital object identifier, HD video and image are at each
Popularize in field and people's daily life.The data volume that high-definition image takies is greatly and the transmission bandwidth in transmission is limited, in future
In foreseeable one period, the resolution data rate of digital picture also will further improve, and existing main flow encoding and decoding
Standard, is such as primarily adapted for use in encoding and decoding standard jpeg, jpeg2000 of still image compression, is primarily adapted for use in video image compression
Encoding and decoding standard mpeg4, h.264 etc., Digital Image Transmission has manifested many deficiencies, in order to adapt to HD video and
The development of Image communicating, needs to develop new Image Compression.
In existing Image Compression, the basic process of encoding and decoding is, for rest image, to be with jpeg standard
Example, input one sub-picture carries out jpeg coding, generates jpeg encoding code stream, then code stream is decoded, and generates decoding and rebuilding
Image；For video, taking h.264 as a example, h.264 input video is encoded, and generates h.264 encoding code stream, then to code stream
H.264 decoded, generated decoding and rebuilding video.The core of jpeg and h.264 standard is conversion.According to signal processing
TIME-FREQUENCY ANALYSIS is theoretical, and the energy in signal is concentrated in a few conversion coefficient, reduces data volume thus can achieve
Compression purpose.From the viewpoint of theory of information, these mainstream scheme be all using pixel in image between dependency it is intended to go
Except statistical redundancy therein.However, through the development of more than 20 years, transform coding method gradually tended to ripe, is not having at this stage
Preferably on the premise of signal analysis means, it to be continued with improvement very difficult.What is more important, HD video figure
As being compressed in data storage and video communication, there is urgent demand and wide application prospect, practice is badly in need of research and is applied to
Video and the new technique of Efficient image compressed encoding.
In Image Compression field, super-resolution rebuilding technology and conventional video and method for encoding images are combined
To obtain the focus that more preferable compression effectiveness is current research.The application for a patent for invention of Application No. 201010292294.1 proposes
A kind of video coding and decoding system based on key frame super-resolution rebuilding, this system is passed through before encoding to video down-sampling,
After the decoding using based on the video super-resolution reconstruction technique with key frame Hierarchical block matching, increase resolution is carried out to video.
The Chen Jingtao of Shanghai Communications University proposes the image compression scheme based on super-resolution rebuilding, and the program is first to low after down-sampling
Image in different resolution carries out encoding and decoding, afterwards decoded low-resolution image is rebuild.The program has also been proposed one simultaneously
Plant the video compress framework of feature based dictionary learning and Kalman filtering estimation, with encoder-side degradation model
Reason partial frame reaches the purpose reducing message complexity, then utilizes reconstruction technique to recover original series in decoder end.
Said method, during by super-resolution rebuilding technology application to video image compression coding, does not take into full account
The quality of decoding image, directly using super-resolution image as final decoding image.For current super-resolution technique, depending on
Registration problems in frequency super-resolution, and single image super-resolution rebuilding lacks the deficiencies such as enough extra prioris, all
Largely limit the quality of super-resolution rebuilding image.Present inventor is tested to said method, from reality
The result tested is seen, these methods compared with traditional encoding and decoding standard, for the video of various resolution, type, in identical figure
As not yet substantially effectively reducing encoder bit rate under quality.Therefore, how preferably super-resolution rebuilding to be fused to compression to compile
In code, in addition to super-resolution rebuilding link, how to improve final decoded image quality under the premise of ensureing compared with low bit- rate is one
Brand-new research work, is shouldering heavy responsibilities of the task that the scientific worker of described technical field faces.
Content of the invention
Resolution data rate with digital picture is constantly lifted, for the importance of image compression encoding, the present invention
Purpose be directed to a kind of new video image compressing method of combination super-resolution rebuilding technology, with realize ensure video
On the premise of picture quality, encoder bit rate is greatly reduced.
The video image compressing method of the combination super-resolution rebuilding that the present invention provides, main include encoding video pictures and
Two parts of video image decoding, described encoding video pictures mainly comprise the steps that
(1) video sequence image of input is carried out down-sampling, obtain the video image after down-sampling；
(2) video image after down-sampling is encoded with encoding and decoding standard, obtain the down-sampled data compressed, use phase
Same encoding and decoding standard is decoded to the down-sampled data of compression, obtains decoded down-sampling video image；
(3) single image is first resolved into decoded down-sampling video image and carry out super-resolution rebuilding again, surpassed
Resolution reconstruction video image；
(4), the video image of input is first resolved into single width original image, then deducted with original single image corresponding
Super-resolution rebuilding single image obtains residual error single image, then residual error single image is synthesized residual video image；
(5) the residual video image obtaining step (4) is encoded with encoding and decoding standard, obtains the residual video compressed
View data；
(6) the residual video image of the compression that the down-sampled data after the compression obtaining step (2) and step (5) obtain
Data merges the data after obtaining inputted video image compression；
Described video image decoding mainly comprises the steps that
(1) the residual error data encoding and decoding standard of the compression in data after compressing inputted video image is decoded,
Then the reconstructive residual error video image obtaining is decomposed into single width residual video image；
(2) the down-sampled data encoding and decoding standard of the compression in data after compressing input picture is decoded, and
Afterwards the reconstructing undersampled video image obtaining first is decomposed into the reconstructing undersampled video image of single width, then adopts under the reconstruction of each frame
Sample video image carries out super-resolution rebuilding, obtains the super-resolution rebuilding video image of down-sampling video image；
(3) step (1) is generated the down-sampling video figure that the single width residual video image obtaining and step (2) generation obtain
The super-resolution rebuilding video image of picture is added, and obtains the reconstruction image of inputted video image.
In the technique scheme of the present invention, described encoding and decoding standard be selected from jpeg, jpeg2000, hevc-i,
H.264 the structure set of criteria with hevc.Particularly, the encoding and decoding standard of rest image preferentially selects jpeg or jpeg2000；Depending on
The encoding and decoding standard of frequency sequence image preferentially selects hevc-i, h.264 or hevc.
In the technique scheme of the present invention, described Downsapling method preferentially selects bicubic algorithm；Described down-sampling
Multiple preferentially elect two times as.
In the technique scheme of the present invention, described super-resolution rebuilding is preferably with theoretical based on sparse signal representation
Super-Resolution of Images Based.
In the described image of technical solution of the present invention encodes, because final encoder bit rate is picture coding code after down-sampling
Final encoder bit rate is therefore defined by rate and residual error auxiliary video and picture coding code check sum in the inventive solutions
For:
bitrateproposed=bitrate1/4+bitrate2
Wherein, bitrate1For down-sampling rear video and picture coding code check in step (2), because the present invention is sampled as two
Times down-sampling, after down-sampling, image is obtained by after two times of down-samplings of original input picture, and its image lattice number is original image 1/4,
Therefore truly code check is bitrate1/4；bitrate2For residual error auxiliary video and picture coding code check in step (5).Due under adopt
Sampled images are different with the robustness of residual image, and respective coded quantization parameter has a significant effect to final code check.So
bitrateproposedValue be according to multiple different quantization parameters of setting respectively in step (2) and step (5), with volume after addition
Decoding standard is compared, code check minimum value preset under identical image Y-PSNR (psnr).Concrete method to set up is:
1st, in step (2), for rest image, using jpeg or jpeg2000 coding, quantization parameter at interval of 10, according to
90～30 setting quantization parameters carry out multiple encoding and decoding respectively, obtain code check when encoding every time, are defined as bitrate1；For
H.264 video sequence image, adopt or hevc encode, quantization parameter carries out multiple encoding and decoding respectively by 28,32,36,40 settings,
Obtain code check when encoding every time, be defined as bitrate1.
2nd, in step (5), for rest image, using in hevc-i coding, quantization parameter at interval of 4, according to 28～40
Setting quantization parameter carries out multiple encoding and decoding respectively, obtains code check when encoding every time, is defined as bitrate2；For video sequence
H.264 row image, adopt or hevc encode, quantization parameter carries out multiple encoding and decoding respectively by 28,32,36,40 settings, obtain every
Code check during secondary coding, is defined as bitrate2.
In the described image of technical solution of the present invention decodes, invention defines following two test index:
δ ssim=ssimproposed-ssimjpeg
Wherein, bitrateproposedAnd bitratejpeg/h.264Represent the method for the present invention, jpeg and h.264 standard respectively
Encoder bit rate, δ bitrate represents that the method for the present invention compares the percentage ratio that jpeg standard code reduces bit rate, for video
Graphical representation compares the percentage ratio that h.264 standard code reduces bit rate；Structural similarity (ssim) is and subjective quality assessment
A kind of stronger picture appraisal standard of relatedness.ssimproposedAnd ssimjpegRepresent the reconstruction image under two methods respectively
Structural similarity value, δ ssim represents that the method for the present invention compares the picture quality of jpeg standard decoding method raising.According to
Method proposed by the present invention, for rest image and jpeg standard, video image, compared with h.264 encryption algorithm, can protected
Hold in the case that image psnr is basically unchanged, greatly save code check.
The present invention traditional jpeg, jpeg2000, h.264 with hevc coding standard algorithm and super resolution image reconstruction skill
Art combines, and devises a kind of new video and method for encoding images.First by original input video image or rest image
Carry out down-sampling, then to image after down-sampling with traditional encoding and decoding standard encoding and decoding, for video image using tradition
H.264 or hevc standard encoding and decoding, jpeg or jpeg2000 standard encoding and decoding are adopted for rest image.Due to down-sampled images
Dot matrix number be only original input picture 1/4, so its encoder bit rate be far below original image encoder bit rate.In side of the present invention
In method, another important step is the super-resolution rebuilding to image after down-sampling.Present invention employs current effect preferably, public
What degree of recognizing was higher carries out super-resolution rebuilding based on the Super-Resolution of Images Based of study to the image after down-sampling.However, it is right
In the super-resolution rebuilding of single image, the prior information due to extracting from low-resolution image is limited after all, so necessarily
Degree limits reconstructed image quality.In order to make up the part detail of the high frequency of loss in super-resolution rebuilding link, carry
Rise reconstructed image quality, on the basis of analysis residual image and the difference of general pattern robustness, design introduces negative inventor
Feedback element, deducts super-resolution rebuilding image with original image and obtains residual image, and this residual image had both contained oversubscription
H.264 or hevc encoding and decoding the high frequency detail lost when resolution is rebuild, carries out hevc-i to this residual image, and residual video carries out
Afterwards, on the super-resolution rebuilding image then before being added back to or video, obtain final reconstruction image.Due to high frequency residual image
Robustness is higher than general pattern, so high frequency residual image can be so that final reconstruction image matter under relatively low encoder bit rate
Amount has a distinct increment.In addition, the code check in the inventive method is saved and is mainly derived from picture coding after down-sampling, and residual image
Be coding side extract and independent of down-sampling after picture coding, so residual video picture coding mode can arbitrarily be selected
Take.
Inventor to the method for the present invention and traditional encoding and decoding standard jpeg, h.264 carried out contrast experiment, experiment knot
Fruit shows, rest image adopts jpeg standard code algorithm, video using h.264 standard code algorithm, in identical image psnr
Under, the method for the present invention greatly reduces encoder bit rate, and to a certain degree improves coded image for parts of images
Subjective visual quality.Concrete effect is shown in the table one of embodiment to table three.
Brief description
Fig. 1 is applied to the encoding and decoding frame diagram of rest image for the present invention.
Fig. 2 is applied to the encoding and decoding frame diagram of video image for the present invention.
Fig. 3 is applied to the encoding and decoding frame diagram of rest image embodiment for the present invention.
Fig. 4 is applied to the encoding and decoding frame diagram of video image embodiment for the present invention.
Fig. 5-1 to Fig. 5-5 is the rest image of experiment test of the present invention.
Fig. 6-1 to Fig. 6-5 is the video image of experiment test of the present invention.
Fig. 7-a to 7-d is the present invention and jpeg, h.264 canonical algorithm rate distortion contrast curve chart, and wherein 7-a is test
Image 5-2 encoding rate distortion curve figure；7-b is test image 5-3 encoding rate distortion curve figure；7-c is test video 6-1 coding
Rate distortion curve figure；7-d is test video 6-5 encoding rate distortion curve figure.
Fig. 8-a and Fig. 8-b is the present invention and jpeg canonical algorithm reconstruction image subjective vision effect contrast figure, wherein 8-a
It is the decoding image of inventive algorithm；8-b is jpeg canonical algorithm decoding image.The object of contrast is test rest image 5-2,
Code check is identical.
Fig. 9-a and Fig. 9-b is the present invention and h.264 canonical algorithm reconstruction image subjective vision effect contrast figure, wherein 9-a
It is the decoding image of inventive algorithm；9-b is that h.264 canonical algorithm decodes image.The object of contrast is test video image 6-
1, code check is identical.
Specific embodiment
With reference to embodiment, the present invention is described in further detail it is necessary to it is noted that below example
It is served only for that the present invention is described further it is impossible to be interpreted as limiting the scope of the invention, art technology is ripe
The personnel that know, according to foregoing invention content, make some nonessential improvement and adjustment are embodied as, should still belong to the present invention
In protection scope of the present invention.
The object of coding is standard still image coding test image, they entitled: Fig. 5-1, Fig. 5-2, Fig. 5-3,
Fig. 5-4, Fig. 5-5, above test image resolution is 512 × 512.The object of Video coding is standard testing video, they
Title, resolution and frame per second be respectively as follows: Fig. 6-1 (352 × 288,30 frames/second), Fig. 6-2 (416 × 240,60 frames/second), figure
6-3 (832 × 480,60 frames/second), Fig. 6-4 (1280 × 720,50 frames/second) and Fig. 6-5 (1920 × 1080,50 frames/second).
For still image coding, using the method with directly adopting jpeg coding for the method for the present invention；For video, adopt
With the method for the present invention with directly encoder bit rate is carried out under identical image psnr using the method h.264 encoding and compares, compare
Process is as follows:
1st, for rest image:
(1) the inventive method encoding and decoding are used to input picture.First input picture is carried out down-sampling, obtain down-sampling
Image.Down-sampling adopts bicubic method, and down-sampling multiple is two times.Secondly, down-sampled images are adopted jpeg or
Jpeg2000 standard carries out encoding and decoding.Then super-resolution rebuilding is carried out to the down-sampled images of decoding.Deducted with input picture
Super-resolution rebuilding image, obtains residual image.Hevc-i encoding and decoding are adopted to residual image, the residual image of decoding with super
Resolution reconstruction image is added and obtains final reconstruction image, and calculates reconstruction image psnr value and ssim value.
(2) qp of a down-sampled images coding, such as qp=30 for down-sampled images encoding and decoding, are set.For residual error
Image carries out hevc-i coding respectively according to qp=28,32,36,40.Final code check is according to calculating defined below:
bitrateproposed=bitrate1/4+bitrate2
Wherein, bitrate1For down-sampled images encoder bit rate, because down-sampled images are under two times of original input picture
Obtain after sampling, its image lattice number is original image 1/4, therefore true code check is bitrate1/4；bitrate2For residual error auxiliary view
As encoder bit rate.Afterwards, change the qp value of down-sampled images coding, residual image is also according to qp=28,32,36,40 respectively
Carry out hevc-i coding.Finally select code check minimum value preset respectively when residual image encodes qp=28,32,36,40 and obtain
To corresponding coded image psnr and ssim value.
(3) jpeg standard encoding and decoding are used to input picture.Adjustment qp value so as to coded image psnr value respectively with process
Four psnr values that 1- (1) obtains are essentially identical (error is ± 0.2db), record corresponding code check respectively.Adjust qp value again, make
Four inventive algorithm code checks that its coded image code check and process 1- (1) obtain are essentially identical (error is ± 0.05kb/s),
Record corresponding ssim value respectively.
(4) method of the present invention is compared with the direct method using jpeg coding, under identical image psnr, uses
Following index is compared to encoder bit rate:
δ ssim=ssimproposed-ssimjpeg
Wherein, bitrateproposedAnd bitratejpegRepresent the present invention and jpeg standard code code check, δ respectively
Bitrate represents that the percentage ratio that jpeg standard code reduces bit rate is compared in coded method of the present invention；ssimproposedWith
ssimjpegRepresent the structural similarity value of the reconstruction image under two methods respectively, δ ssim represents decoding method of the present invention
The picture quality that comparison with standard jpeg decoding method improves.
2nd, for video image:
(1) the inventive method encoding and decoding are used to input video.First inputted video image is carried out down-sampling, under obtaining
Sample video image.Down-sampling adopts bicubic method, and down-sampling multiple is two times.Secondly, down-sampling video image is adopted
Or hevc standard carries out encoding and decoding h.264.Then the down-sampling video image of decoding is decomposed into single image, to each image
Carry out super-resolution rebuilding.Meanwhile, inputted video image is decomposed into single image, deducts corresponding super-resolution with input picture
Rate reconstruction image, obtains residual image.Single width residual image is synthesized residual video image, residual video image is adopted
H.264 or hevc encoding and decoding, it is single image phase corresponding with super-resolution rebuilding image the residual video picture breakdown of decoding
Plus obtain final reconstruction image, finally single width reconstruction image is synthesized reconstruction video image, and calculate reconstruction video psnr value.
(2) qp of down-sampling encoding video pictures, such as qp=24 to down-sampling video image encoding and decoding, are set.Right
Carry out respectively h.264 according to qp=28,32,36,40 in residual video image or hevc encodes.Final code check is according to defined below
Calculate:
bitrateproposed=bitrate1/4+bitrate2
Wherein, bitrate1For down-sampling encoding video pictures code check, because down-sampled images are original input picture two
Obtain after times down-sampling, its image lattice number is original image 1/4, therefore true code check is bitrate1/4；bitrate2Auxiliary for residual error
Help encoding video pictures code check.Afterwards, change the qp value of down-sampling encoding video pictures, residual video image is also according to qp=
28th, 32,36,40 carry out respectively h.264 or hevc coding.Final respectively residual video picture coding qp=28,32,36,40
When select code check minima and obtain corresponding encoded video image psnr value.
(3) h.264 standard encoding and decoding are used to inputted video image.Adjustment qp value is so as to coded image psnr value is distinguished
Essentially identical with four psnr values that process 2- (1) obtains (error is ± 0.2db), records corresponding code check respectively.
(4) by the method for the present invention be directly compared using the method h.264 encoding, under identical image psnr,
With following index, encoder bit rate is compared:
Wherein, bitrateproposedAnd bitrateh.264The expression present invention and h.264 standard code code check respectively, δ
Bitrate represents that coding framework of the present invention compares the percentage ratio that h.264 standard code reduces bit rate.
The result of above-mentioned 3 indexs is as shown in table 1 is to table 3.Wherein, table 1 be for rest image the inventive method with
The comparison of jpeg standard encoder bit rate under identical coded image psnr, table 2 is for rest image the inventive method and jpeg
The comparison of standard ssim value under identical encoder bit rate, table 3 is in identical coding for video the inventive method with h.264 standard
The comparison of encoder bit rate under image psnr.Statistics shows the inventive method compared with jpeg standard encoding and decoding, in identical image
Under psnr, encoder bit rate averagely reduces by 29.88%；Compared with h.264 standard encoding and decoding, under identical image psnr, encode code
Rate averagely reduces by 50.54%, greatly reduces encoder bit rate.Meanwhile, inventive algorithm is to reconstruction image subjective vision
Quality has a certain degree of lifting.
The comparison of table 1 the inventive method and jpeg standard encoder bit rate under identical coded image psnr
The comparison of table 2 the inventive method and jpeg standard ssim value under identical encoder bit rate
The comparison of table 3 the inventive method and h.264 standard encoder bit rate under identical coded image psnr
Claims (9)
1. a kind of video image compressing method of combination super-resolution rebuilding it is characterised in that mainly include encoding video pictures and
Two parts of video image decoding, described encoding video pictures mainly comprise the steps that
(1) video sequence image of input is carried out down-sampling, obtain the video image after down-sampling；
(2) video image after down-sampling is encoded with encoding and decoding standard, obtain the down-sampled data compressed, use identical
Encoding and decoding standard is decoded to the down-sampled data of compression, obtains decoded down-sampling video image；
(3) single image is first resolved into decoded down-sampling video image and carry out super-resolution rebuilding again, obtain super-resolution
Rate rebuilds video image；
(4) video image of input is first resolved into single width original image, then deduct corresponding oversubscription with original single image
Resolution is rebuild single image and is obtained residual error single image, then residual error single image is synthesized residual video image；
(5) the residual video image obtaining step (4) is encoded with encoding and decoding standard, obtains the residual video image compressing
Data；
(6) the residual video view data of the compression that the down-sampled data after the compression obtaining step (2) and step (5) obtain
Merge the data after obtaining inputted video image compression；
Described video image decoding mainly comprises the steps that
(1) the residual error data encoding and decoding standard of the compression in data after compressing inputted video image is decoded, then
The reconstructive residual error video image obtaining is decomposed into single width residual video image；
(2) the down-sampled data encoding and decoding standard of the compression in data after compressing input picture is decoded, then
The reconstructing undersampled video image obtaining first is decomposed into the reconstructing undersampled video image of single width, more reconstructing undersampled to each frame regards
Frequency image carries out super-resolution rebuilding, obtains the super-resolution rebuilding video image of down-sampling video image；
(3) step (1) is generated the single width residual video image obtaining and step (2) generates the down-sampling video image obtaining
Super-resolution rebuilding video image is added, and obtains the reconstruction image of inputted video image.
2. the method for compressing image of combination super-resolution rebuilding according to claim 1 is it is characterised in that described encoding and decoding
Standard is selected from jpeg, jpeg2000, hevc-i, h.264 the structure set of criteria with hevc.
3. the method for compressing image of combination super-resolution rebuilding according to claim 2 is it is characterised in that to down-sampling
Image adopts in jpeg or jpeg2000 standard code, and quantization parameter is spaced apart 10, according to 90～30 setting quantization parameters respectively
Carry out multiple encoding and decoding.
4. the method for compressing image of combination super-resolution rebuilding according to claim 2 is it is characterised in that to residual plot
As using in hevc-i standard code, quantization parameter is spaced apart 4, repeatedly compiled respectively according to 28～40 setting quantization parameters
Decoding.
5. the method for compressing image of combination super-resolution rebuilding according to claim 2 is it is characterised in that to residual plot
As using in hevc standard code, quantization parameter carries out multiple encoding and decoding respectively by 28,32,36,40 settings.
6. the method for compressing image of the combination super-resolution rebuilding according to one of claim 1 to 5 is it is characterised in that described
Downsapling method selects bicubic algorithm, and down-sampling multiple is two times.
7. the method for compressing image of the combination super-resolution rebuilding according to one of claim 1 to 5 is it is characterised in that surpassing
Resolution reconstruction is using the Super-Resolution of Images Based theoretical based on sparse signal representation.
8. the method for compressing image of combination super-resolution rebuilding according to claim 7 is it is characterised in that in super-resolution
In process of reconstruction, training dictionary is generated according to low-resolution image and Its Sparse Decomposition algorithm.
9. the method for compressing image of combination super-resolution rebuilding according to claim 8 was it is characterised in that described trained
The generation of journey and dictionary is to complete before encoding and decoding.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310518474.0A CN103607591B (en) | 2013-10-28 | 2013-10-28 | Video image compression method combining super-resolution reconstruction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310518474.0A CN103607591B (en) | 2013-10-28 | 2013-10-28 | Video image compression method combining super-resolution reconstruction |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103607591A CN103607591A (en) | 2014-02-26 |
| CN103607591B true CN103607591B (en) | 2017-02-01 |
Family
ID=50125788
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310518474.0A Active CN103607591B (en) | 2013-10-28 | 2013-10-28 | Video image compression method combining super-resolution reconstruction |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103607591B (en) |
Families Citing this family (26)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105392009B (en) * | 2015-11-27 | 2019-04-16 | 四川大学 | Low bit rate image sequence coding method based on block adaptive sampling and super-resolution rebuilding |
| CN105959705B (en) * | 2016-05-10 | 2018-11-13 | 武汉大学 | A kind of net cast method towards wearable device |
| CN106251373B (en) * | 2016-07-19 | 2019-06-25 | 四川大学 | A kind of image compression encoding method merging super-resolution technique and JPEG2000 standard |
| CN106961610B (en) * | 2017-03-15 | 2020-05-05 | 四川大学 | Novel super-high-definition video compression framework combined with super-resolution reconstruction |
| CN107181949A (en) * | 2017-06-23 | 2017-09-19 | 四川大学 | A kind of compression of images framework of combination super-resolution and residual coding technology |
| CN109218727B (en) * | 2017-06-30 | 2021-06-25 | 书法报视频媒体(湖北)有限公司 | Video processing method and device |
| CN109547784A (en) * | 2017-09-21 | 2019-03-29 | 华为技术有限公司 | A kind of coding, coding/decoding method and device |
| US11082720B2 (en) | 2017-11-21 | 2021-08-03 | Nvidia Corporation | Using residual video data resulting from a compression of original video data to improve a decompression of the original video data |
| CN109905717A (en) * | 2017-12-11 | 2019-06-18 | 四川大学 | A kind of H.264/AVC Encoding Optimization based on Space-time domain down-sampling and reconstruction |
| CN109922339A (en) * | 2017-12-12 | 2019-06-21 | 四川大学 | In conjunction with the image coding framework of multi-sampling rate down-sampling and super-resolution rebuilding technology |
| CN108737823B (en) * | 2018-04-04 | 2021-05-25 | 中国传媒大学 | Image coding method and device and decoding method and device based on super-resolution technology |
| CN109214989B (en) * | 2018-09-04 | 2019-08-13 | 四川大学 | Single image super resolution ratio reconstruction method based on Orientation Features prediction priori |
| CN109451323A (en) * | 2018-12-14 | 2019-03-08 | 上海国茂数字技术有限公司 | A kind of lossless image coding method and device |
| CN111510739B (en) * | 2019-01-31 | 2022-04-29 | 华为技术有限公司 | Video transmission method and device |
| CN111565317A (en) * | 2019-02-13 | 2020-08-21 | 合肥图鸭信息科技有限公司 | Image compression method, coding and decoding network training method and device and electronic equipment |
| CN110062232A (en) * | 2019-04-01 | 2019-07-26 | 杭州电子科技大学 | A kind of video-frequency compression method and system based on super-resolution |
| CN111800629A (en) * | 2019-04-09 | 2020-10-20 | 华为技术有限公司 | Video decoding method, video encoding method, video decoder and video encoder |
| CN112188237A (en) * | 2019-07-04 | 2021-01-05 | 国家广播电视总局广播电视科学研究院 | Program distribution method and device, receiving method, terminal device and medium |
| CN111343458B (en) * | 2020-01-15 | 2022-03-01 | 浙江工业大学 | Sparse gray image coding and decoding method and system based on reconstructed residual |
| CN113810763A (en) * | 2020-06-15 | 2021-12-17 | 深圳市中兴微电子技术有限公司 | Video processing method, device and storage medium |
| CN111754406B (en) * | 2020-06-22 | 2024-02-23 | 北京大学深圳研究生院 | Image resolution processing method, device, equipment and readable storage medium |
| CN111898638B (en) * | 2020-06-29 | 2022-12-02 | 北京大学 | Image processing method, electronic device and medium fusing different visual tasks |
| CN112233197B (en) * | 2020-09-14 | 2024-03-26 | 北京空间飞行器总体设计部 | Compression coding method for spacecraft on-orbit image |
| CN113014927B (en) * | 2021-03-02 | 2024-01-09 | 三星（中国）半导体有限公司 | Image compression method and image compression device |
| CN113240107A (en) * | 2021-05-08 | 2021-08-10 | 北京字跳网络技术有限公司 | Image processing method and device and electronic equipment |
| CN113938684B (en) * | 2021-09-26 | 2023-04-21 | 中国科学院计算技术研究所 | Lightweight end-cloud collaborative image coding transmission method and system |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101938656A (en) * | 2010-09-27 | 2011-01-05 | 上海交通大学 | Video coding and decoding system based on keyframe super-resolution reconstruction |
| CN102986220A (en) * | 2010-07-20 | 2013-03-20 | 西门子公司 | Video coding with reference frames of high resolution |
-
2013
- 2013-10-28 CN CN201310518474.0A patent/CN103607591B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102986220A (en) * | 2010-07-20 | 2013-03-20 | 西门子公司 | Video coding with reference frames of high resolution |
| CN101938656A (en) * | 2010-09-27 | 2011-01-05 | 上海交通大学 | Video coding and decoding system based on keyframe super-resolution reconstruction |
Non-Patent Citations (1)
| Title |
|---|
| 《基于超分辨率重建方法的图像视频压缩编码技术研究》;陈静涛;《中国优秀硕士学位论文全文数据库信息科技辑》;20120731(第7期);第3.1-3.2节，图3-1 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103607591A (en) | 2014-02-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103607591B (en) | Video image compression method combining super-resolution reconstruction | |
| CN107018422B (en) | Still image compression method based on depth convolutional neural networks | |
| Hu et al. | FVC: A new framework towards deep video compression in feature space | |
| CN107197260B (en) | Video coding post-filter method based on convolutional neural networks | |
| CN109495741B (en) | Image compression method based on self-adaptive down-sampling and deep learning | |
| CN101742313B (en) | Compression sensing technology-based method for distributed type information source coding | |
| CN110087092A (en) | Low bit-rate video decoding method based on image reconstruction convolutional neural networks | |
| CN110751597B (en) | Video super-resolution method based on coding damage repair | |
| CN104065976B (en) | A kind of compression of images and security transmission method based on video | |
| CN101938656A (en) | Video coding and decoding system based on keyframe super-resolution reconstruction | |
| CN105392009B (en) | Low bit rate image sequence coding method based on block adaptive sampling and super-resolution rebuilding | |
| CN113205456A (en) | Super-resolution reconstruction method for real-time video session service | |
| CN109361919A (en) | A kind of image coding efficiency method for improving combined super-resolution and remove pinch effect | |
| CN109922339A (en) | In conjunction with the image coding framework of multi-sampling rate down-sampling and super-resolution rebuilding technology | |
| CN106961610A (en) | With reference to the ultra high-definition video new type of compression framework of super-resolution rebuilding | |
| CN110099280A (en) | A kind of video service quality Enhancement Method under wireless self-organization network Bandwidth-Constrained | |
| CN111711817A (en) | HEVC intra-frame coding compression performance optimization research combined with convolutional neural network | |
| CN104780383B (en) | A kind of 3D HEVC multi-resolution video coding methods | |
| CN110545426B (en) | Spatial domain scalable video coding method based on coding damage repair (CNN) | |
| CN105825530B (en) | Littoral zone high spectrum image distribution lossy coding and coding/decoding method based on area-of-interest | |
| CN111726638A (en) | HEVC (high efficiency video coding) optimization method combining decompression effect and super-resolution | |
| CN106534870B (en) | A kind of rate-distortion optimal coding method based on RGB source video | |
| CN115442613A (en) | Interframe information-based noise removal method using GAN | |
| CN106251373B (en) | A kind of image compression encoding method merging super-resolution technique and JPEG2000 standard | |
| Tan et al. | Image compression algorithms based on super-resolution reconstruction technology |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant |