CN108650509B - Multi-scale self-adaptive approximate lossless coding and decoding method and system - Google Patents
Multi-scale self-adaptive approximate lossless coding and decoding method and system Download PDFInfo
- Publication number
- CN108650509B CN108650509B CN201810293916.9A CN201810293916A CN108650509B CN 108650509 B CN108650509 B CN 108650509B CN 201810293916 A CN201810293916 A CN 201810293916A CN 108650509 B CN108650509 B CN 108650509B
- Authority
- CN
- China
- Prior art keywords
- detail
- trend
- original image
- image
- encoder
- 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
- 238000000034 method Methods 0.000 title claims abstract description 31
- 230000003044 adaptive effect Effects 0.000 claims abstract description 7
- 238000012549 training Methods 0.000 claims description 7
- 239000000284 extract Substances 0.000 claims description 4
- 238000007906 compression Methods 0.000 abstract description 10
- 230000006835 compression Effects 0.000 abstract description 9
- 230000008569 process Effects 0.000 description 4
- 238000010586 diagram Methods 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 2
- 230000006837 decompression Effects 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 230000005484 gravity Effects 0.000 description 2
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 201000010099 disease Diseases 0.000 description 1
- 208000037265 diseases, disorders, signs and symptoms Diseases 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000010354 integration Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000000452 restraining effect Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 230000037303 wrinkles Effects 0.000 description 1
Images
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
- H04N19/136—Incoming video signal characteristics or properties
- H04N19/14—Coding unit complexity, e.g. amount of activity or edge presence estimation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
-
- 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
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/124—Quantisation
-
- 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
- H04N19/146—Data rate or code amount at the encoder output
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Compression Of Band Width Or Redundancy In Fax (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
A multi-scale self-adaptive approximate lossless coding and decoding method comprises the following steps: 1) dividing an original image into a plurality of scale levels with different sizes, and extracting image features under different scales by using a multi-scale feature extractor; 2) the image features under different scales are coded by a detail coder to obtain detail hidden variables of the original image; 3) the trend encoder encodes the last layer of image characteristics to obtain a trend encoding hidden variable of the original image, and the trend encoding hidden variable and the detail hidden variable jointly form approximate lossless encoding of the original image; 4) decoding the trend hidden variable to obtain a fuzzy contour of the original image; 5) decoding the detail hidden variable to obtain the detail information of the original image; 6) and accumulating the decoded fuzzy contour and the detail information to obtain the approximate lossless restoration of the original image. And provides a multi-scale adaptive near-lossless coding and decoding system. The invention improves the compression ratio of the image and solves the problem of detail loss in the image compression process.
Description
Technical Field
The invention belongs to the field of image coding and decoding, and particularly relates to a method and a system for adaptively extracting image characteristic information on different scales for image approximate lossless coding and decoding.
Background
With the development of computer technology, computer images are widely used in various fields. In many fields, there is a need to use high-definition computer images that contain rich detailed information. For example, CT pictures in the medical field, detailed information such as a small white dot is also an important reference for doctors to diagnose the disease; for example, in the field of aircraft wing design, the detail information of the wing edge can have great influence on the aerodynamics of the wing; for example, in the field of face recognition, details of a face image such as spots or fine wrinkles on the face may also affect the accuracy of face recognition.
Raw computer image data that is not processed typically contains color and intensity information for each pixel in the image. High precision images typically contain millions or even tens of millions of pixels, resulting in an excessive amount of storage space for the raw image that is not compressed.
The coding and decoding of the image, also called image compression and decompression, carries out compression coding on the image by finding and recording the rule among pixels, and realizes the restoration of the original image through a matched decompression algorithm, thereby realizing the reduction of the storage space occupied by the image.
Larson et al propose a variational self-encoder and a codec structure for generating a countermeasure network, wherein an encoder maps an input image into a hidden variable, a decoder is also a generator in the generation countermeasure network and is responsible for decoding the hidden variable into an image, and the decoder needs to minimize an image reconstruction penalty function and a judger function at the same time. The decoupling learning method for the conditional countermeasure network provides a training method for improving the learning stability of the self-coding generation countermeasure network, can reduce the training difficulty of the network and improve the precision of a decoding result, but the experimental network structure is simpler, and the used method has larger promotion space.
In the method disclosed in patent CN102480620, the compression of the image is realized by recording the frequency characteristics of the image by an integration method, but the method only uses a single frequency characteristic, and has the inevitable problem of detail loss.
In the method disclosed in CN104349171, a visually lossless image codec is implemented by dividing an image into several macro blocks, analyzing and classifying each macro block, and performing different compression algorithms for different macro block categories. However, the threshold parameter used in the method for classifying the macro block has a large influence on the accuracy and quality of the restored image, and it is difficult to obtain the optimal threshold value for classifying the macro block. And for pictures of different application scenes, the threshold used by macroblock classification is often different, which requires a long time to adjust the threshold parameter.
Disclosure of Invention
In order to overcome the defect that the details of the existing coding and decoding method are easy to lose, the invention provides an approximately lossless coding and decoding method and system, which are used for solving the problem that the details are lost in picture compression.
The technical scheme adopted by the invention for solving the technical problems is as follows:
a multi-scale adaptive near-lossless coding and decoding method, the method comprising the steps of:
1) dividing an original image into a plurality of scale levels with different sizes, and extracting image features under different scales by using a multi-scale feature extractor;
2) the image features under different scales are coded by a detail coder to obtain detail hidden variables of the original image;
3) the trend encoder encodes the last layer of image characteristics to obtain a trend encoding hidden variable of the original image, and the trend encoding hidden variable and the detail hidden variable jointly form approximate lossless encoding of the original image;
4) and decoding the trend hidden variable by a trend decoder to obtain the fuzzy contour of the original image.
5) Decoding the detail hidden variable by a detail decoder to obtain detail information of the original image;
6) and accumulating the decoded fuzzy contour and the detail information to obtain the approximate lossless restoration of the original image.
A multi-scale adaptive near-lossless codec system, the system comprising: the hidden variable coder is used for compressing the high-dimensional data information into low-dimensional coding information; and the approximate lossless reconstruction module is used for approximately losslessly restoring the low-dimensional decoding information into data before encoding.
Further, the hidden variable encoder includes: multi-scale feature extractor, usingExtracting the features of the original image under different scales; a detail encoder O for generating detail hidden variables in different scales, denoted as E in FIG. 1all(ii) a A trend encoder V for generating a trend hidden variable (E in FIG. 1) containing the fuzzy contour information of the original image9。
In the detail encoder O, the hidden variables obey some predefined prior probability distribution, e.g. a uniform random distribution.
In the detail encoder O, the detail information and the specific gravity included in the encoding are adaptively adjusted according to the statistical characteristics of the training set.
The multi-scale feature extractor extracts features from an input image in a multi-scale layer-by-layer manner to obtain a plurality of feature maps, and the feature maps are different in size.
And in the trend encoder V, the image features extracted at the last layer by the multi-scale feature extractor are compressed and encoded into trend hidden variables.
Still further, the lossless reconstruction module includes: the trend decoder D restores the fuzzy contour of the original image according to the trend hidden variable generated by the trend encoder; the detail decoder G restores the detail information of the original image according to the detail hidden variables generated by the detail encoder; and the decoding synthesizer comprehensively analyzes and processes the decoding results generated by the trend decoder and the detail decoder to approximately restore the original image without loss.
The number of detail decoders G is equal to the number of detail decoders O; the number of trend decoders D is equal to the number of trend encoders V.
Each detail decoder G can restore partial details of the original image, and the output results of the detail decoders together form the complete detail information of the original image.
The invention has the following beneficial effects: 1, introducing a Laplacian pyramid structure to extract the characteristics of an original image under different scales, and highly retaining the detail information of the image; 2. introducing and generating a spatial constraint of the confrontation network on the encoded detail information to ensure that the confrontation network conforms to a certain predefined probability distribution; 3. the encoder can adaptively distribute the detail information contained in the characteristic diagram under each scale through the abstract process of adaptively adjusting the abstract process to the input image according to the statistical characteristics of the training database.
Drawings
FIG. 1 is a schematic diagram of the main structure of the multi-scale adaptive lossless codec system according to the present invention.
FIG. 2 is a schematic structural diagram of an example of the multi-scale adaptive lossless codec method according to the present invention.
Original reference number, X represents the original image to be coded and decoded, Y represents the restored image after coding and decoding, F0～F9Representing a multi-scale feature, N1～N9Number of feature maps representing the correspondence of the above features, E0～E8Representing a sub-encoder, DL0～DL8Representing a hidden variable determiner, DB representing a reconstructed image determiner, D representing a trend decoder, G0～G9Representing a sub-section encoder, V1～V9Representing a multi-scale feature extractor, O1～O9A multi-scale information filter is shown, and ys shows the trend reconstruction result of the original image x.
Detailed Description
The invention is further described below with reference to the accompanying drawings.
Referring to fig. 1, a multi-scale adaptive near-lossless codec system, the system comprising: the hidden variable coder is used for compressing the high-dimensional data information into low-dimensional coding information; and the approximate lossless reconstruction module is used for approximately losslessly restoring the low-dimensional decoding information into data before encoding.
Further, the hidden variable encoder includes: the multi-scale feature extractor is used for extracting features of the original image under different scales; a detail encoder O for generating detail hidden variables in different scales, denoted as E in FIG. 1all(ii) a A trend encoder V for generating a trend hidden variable (E in FIG. 1) containing the fuzzy contour information of the original image9。
In the detail encoder O, the hidden variables obey some predefined prior probability distribution, e.g. a uniform random distribution.
In the detail encoder O, the detail information and the specific gravity included in the encoding are adaptively adjusted according to the statistical characteristics of the training set.
The multi-scale feature extractor extracts features from an input image in a multi-scale layer-by-layer manner to obtain a plurality of feature maps, and the feature maps are different in size.
And in the trend encoder V, the image features extracted at the last layer by the multi-scale feature extractor are compressed and encoded into trend hidden variables.
Still further, the lossless reconstruction module includes: the trend decoder D restores the fuzzy contour of the original image according to the trend hidden variable generated by the trend encoder; the detail decoder G restores the detail information of the original image according to the detail hidden variables generated by the detail encoder; and the decoding synthesizer comprehensively analyzes and processes the decoding results generated by the trend decoder and the detail decoder to approximately restore the original image without loss.
The number of detail decoders G is equal to the number of detail decoders O; the number of trend decoders D is equal to the number of trend encoders V.
Each detail decoder G can restore partial details of the original image, and the output results of the detail decoders together form the complete detail information of the original image.
Referring to fig. 2, the near lossless compression coding/decoding method of an embodiment includes the following steps:
1) dividing the original image into n scale levels of different sizes, and using a feature extractor ViThe feature map extracted from each level and the original image together form a feature set F0,F1,…,Fn；
2) Layer i characteristic FiVia information filter OiThe processed result and the previous layer characteristic Fi-1Add, and then go through a sub-encoder EiObtaining a set of implicit variables z after encoding0,z1,..,zn. And through an implicit variable decider DLiTo constrain the hidden variable ziAm (a)Rate distribution;
3) for implicit variable sets, a sub-detail decoder G is usediRestoring detail information of the image under different scales;
4) for the last layer coding result znUsing a trend decoder D to restore and obtain fuzzy information of the original image;
5) and accumulating the decoding results of the sub-section decoder and the decoding results of the trend decoder to realize the approximately lossless restoration of the original image.
To the left of the hidden variable z in fig. 2 is an encoder of the approximate lossless compression coding and decoding method, the encoder comprising:
multi-scale feature extractor V1,V2,…,V9And the method is used for extracting detail features of the images under different scales. The feature graphs after feature extraction are respectively F1,F2,…,F9. Furthermore, N1,N2,…,N9Respectively the number of feature maps at different scales.
In some embodiments, the feature extractor may extract features of the image using a convolutional neural network. It should be understood, however, that the multi-scale feature extractor of the present invention is not limited to extracting image features using convolutional neural networks, and the method for extracting features may be different at each scale.
Multi-scale information filter O1,O2,…,O9And the method is used for screening the detail information of the feature maps under different scales, and adding the screened feature information and the feature map under the previous scale.
Referring to fig. 2, after the feature map F is processed by the information filter O, the dimension of the feature map F is the same as that of the feature map of the previous layer, so that a mathematical addition operation can be performed.
Sub-encoder EiCoding the summation of the characteristic graphs under different scales to obtain an implicit variable z after codingi。
In a specific application, a user can adjust the dimension of the hidden variable z according to actual requirements.
DLiTo be corresponding toVariable ziFor constraining ziFitting it to some predefined probability distribution.
Hidden variable z in FIG. 2iTo the right of (2) is a decoder of the approximate lossless compression codec method, the decoder comprising:
Gifor a sub-detail decoder, for decoding an implicit variable z containing image detail information at different scalesi。
D is a trend decoder for decoding the last layer of feature map z9The decoded image ys is the trend reconstruction result of the original image x.
And DB is a reconstruction decision device and is used for restraining the error between the image y after encoding and decoding and the image x before encoding and decoding.
That is, when the result output by the reconstruction determiner is smaller than a certain threshold, the reconstructed image after encoding and decoding can be considered to meet the expected requirement.
In this embodiment, the input image x has a length and width of 1024 and the number of channels is 1. In some embodiments, the length and width of the image may be greater or less than 1024, and the number of channels may be more than 1.
Referring to FIG. 2, this embodiment uses a total of 9 feature extractors V1,V2,…,V9,. In some embodiments, the number of multi-scale feature extractors V may not be equal to the number of multi-scale feature extractors V in the present embodiment.
While the embodiments of the present invention have been described with reference to specific examples, those skilled in the art will readily appreciate still other advantages and features of the invention from the disclosure herein. The invention may be embodied or carried out in various other specific forms, and it is to be understood that various changes, modifications, and alterations may be made in the details of the description without departing from the spirit of the invention. It should be noted that the features in the following embodiments and examples may be combined with each other without conflict.
It should be noted that the drawings provided in the above embodiments are only for illustrating the basic idea of the present invention, and the components related to the present invention are only shown in the drawings rather than drawn according to the number of components in actual implementation, and the number and the proportion of the components in actual implementation can be changed freely.
Claims (2)
1. A multi-scale self-adaptive approximate lossless coding and decoding method is characterized in that: the method comprises the following steps:
1) dividing an original image into a plurality of scale levels with different sizes, and extracting image features under different scales by using a multi-scale feature extractor;
2) the image features under different scales are coded by a detail coder to obtain detail hidden variables of the original image;
3) the trend encoder encodes the image features extracted by the multi-scale feature extractor at the last layer to obtain a trend encoding hidden variable of the original image, and the trend encoding hidden variable and the detail hidden variable jointly form an approximate lossless encoding of the original image;
in the detail encoder and the trend encoder, hidden variables obey a certain predefined prior probability distribution; in the detail encoder, the detail information and the proportion contained in the encoding are adaptively adjusted according to the statistical characteristics of the training set;
4) decoding the trend hidden variable by a trend decoder to obtain a fuzzy profile of the original image;
5) decoding the detail hidden variable by a detail decoder to obtain detail information of the original image;
6) and accumulating the decoded fuzzy contour and the detail information to obtain the approximate lossless restoration of the original image.
2. A system implemented by the multi-scale adaptive near lossless coding and decoding method according to claim 1, wherein the system comprises: a hidden variable encoder for compressing high-dimensional image data into low-dimensional encoding information; the approximate lossless reconstruction module is used for restoring the low-dimensional decoding information into an image before encoding in an approximately lossless manner;
the latent variable encoder includes: a multi-scale feature extractor for extracting original imageLike features at different scales; a detail encoder O for generating detail hidden variables E under different scalesall(ii) a A trend encoder V for generating a trend hidden variable E containing the fuzzy contour information of the original image9(ii) a The hidden variables generated by the detail encoder O and the trend encoder V obey a certain predefined prior probability distribution; the detail encoder O adaptively adjusts detail information and proportion contained in the encoding according to the statistical characteristics of the training set; the multi-scale feature extractor extracts features from an input image in a multi-scale layer-by-layer manner to obtain a plurality of feature maps, and the feature maps are different in size; the method comprises the steps that the features extracted by a multi-scale feature extractor on the last layer are regarded as the features containing the original image fuzzy contour, and a trend encoder V is used for compressing and encoding the features of the last layer into trend hidden variables;
the near lossless reconstruction module includes: the trend decoder D restores the fuzzy contour of the original image according to the trend hidden variable generated by the trend encoder; the detail decoder G restores the detail information of the original image according to the detail hidden variables generated by the detail encoder; the decoding synthesizer is used for comprehensively analyzing and processing decoding results generated by the trend decoder and the detail decoder to approximately restore an original image in a lossless manner; the number of detail decoders G is equal to the number of detail encoders O; the number of the trend decoders D is equal to the number of the trend encoders V; each detail decoder G can restore partial details of the original image, and the output results of the detail decoders together form the complete detail information of the original image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810293916.9A CN108650509B (en) | 2018-04-04 | 2018-04-04 | Multi-scale self-adaptive approximate lossless coding and decoding method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810293916.9A CN108650509B (en) | 2018-04-04 | 2018-04-04 | Multi-scale self-adaptive approximate lossless coding and decoding method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108650509A CN108650509A (en) | 2018-10-12 |
| CN108650509B true CN108650509B (en) | 2020-08-18 |
Family
ID=63745250
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810293916.9A Active CN108650509B (en) | 2018-04-04 | 2018-04-04 | Multi-scale self-adaptive approximate lossless coding and decoding method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108650509B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111224670B (en) * | 2018-11-27 | 2023-09-15 | 富士通株式会社 | Automatic encoder, and method and medium for training the same |
| CN113012052B (en) * | 2019-12-19 | 2022-09-20 | 浙江商汤科技开发有限公司 | Image processing method and device, electronic equipment and storage medium |
| CN111695598B (en) * | 2020-05-11 | 2022-04-29 | 东南大学 | Monitoring data abnormity diagnosis method based on generation countermeasure network |
| CN112597702B (en) * | 2020-12-21 | 2022-07-19 | 电子科技大学 | Pneumatic modeling generation type confrontation network model training method based on radial basis function |
| CN114663536B (en) * | 2022-02-08 | 2022-12-06 | 中国科学院自动化研究所 | Image compression method and device |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103985105A (en) * | 2014-02-20 | 2014-08-13 | 江南大学 | Contourlet domain multi-modal medical image fusion method based on statistical modeling |
| CN106339717A (en) * | 2016-08-17 | 2017-01-18 | 衢州学院 | Living standing tree rapid identification method based on hybrid multi-scale feature model |
| CN107341786A (en) * | 2017-06-20 | 2017-11-10 | 西北工业大学 | The infrared and visible light image fusion method that wavelet transformation represents with joint sparse |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11221990B2 (en) * | 2015-04-03 | 2022-01-11 | The Mitre Corporation | Ultra-high compression of images based on deep learning |
-
2018
- 2018-04-04 CN CN201810293916.9A patent/CN108650509B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103985105A (en) * | 2014-02-20 | 2014-08-13 | 江南大学 | Contourlet domain multi-modal medical image fusion method based on statistical modeling |
| CN106339717A (en) * | 2016-08-17 | 2017-01-18 | 衢州学院 | Living standing tree rapid identification method based on hybrid multi-scale feature model |
| CN107341786A (en) * | 2017-06-20 | 2017-11-10 | 西北工业大学 | The infrared and visible light image fusion method that wavelet transformation represents with joint sparse |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108650509A (en) | 2018-10-12 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108650509B (en) | Multi-scale self-adaptive approximate lossless coding and decoding method and system | |
| CN113259676B (en) | Image compression method and device based on deep learning | |
| Wu et al. | A gan-based tunable image compression system | |
| CN111149363A (en) | Information processing apparatus and method | |
| CN111108529A (en) | Information processing apparatus and method | |
| US11893762B2 (en) | Method and data processing system for lossy image or video encoding, transmission and decoding | |
| CN111246206A (en) | Optical flow information compression method and device based on self-encoder | |
| CN113947136A (en) | Image compression and classification method and device and electronic equipment | |
| US20080226187A1 (en) | Image coding method, image coding apparatus, image coding circuit, information recording medium, and computer program | |
| Otair et al. | Improved near-lossless technique using the Huffman coding for enhancing the quality of image compression | |
| Ranjbar Alvar et al. | Joint image compression and denoising via latent-space scalability | |
| Zepeda et al. | Image compression using the iteration-tuned and aligned dictionary | |
| CN111343458B (en) | Sparse gray image coding and decoding method and system based on reconstructed residual | |
| Akbari et al. | Image compression using adaptive sparse representations over trained dictionaries | |
| Garg et al. | Analysis of different image compression techniques: a review | |
| CN115049055B (en) | Super-division neural network quantification method based on dynamic double trainable limit | |
| Akbari et al. | Downsampling based image coding using dual dictionary learning and sparse representations | |
| Thakker et al. | Lossy Image Compression-A Comparison Between Wavelet Transform, Principal Component Analysis, K-Means and Autoencoders | |
| George et al. | A novel approach for image compression using zero tree coding | |
| CN115361559A (en) | Image encoding method, image decoding method, image encoding device, image decoding device, and storage medium | |
| Naaz et al. | Implementation of hybrid algorithm for image compression and decompression | |
| Küchhold et al. | Restricted Boltzmann machine image compression | |
| Garg et al. | Various Image Compression Techniques: A Review. | |
| CN117714697B (en) | Digital human video display method and device | |
| US20130279804A1 (en) | Dual transform lossy and lossless compression |
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 | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20231030 Address after: Room 301-1, Building 2, Academician Island, Xianghu National Tourism Resort, Xiaoshan District, Hangzhou City, Zhejiang Province, 311258 Patentee after: Hangzhou Jiuwei Digital Technology Co.,Ltd. Address before: The city Zhaohui six districts Chao Wang Road Hangzhou City, Zhejiang province 310014 18 Patentee before: JIANG University OF TECHNOLOGY |