CN108419080B - Method and device for streamline optimization of JPEGLS context calculation - Google Patents
Method and device for streamline optimization of JPEGLS context calculation Download PDFInfo
- Publication number
- CN108419080B CN108419080B CN201810130397.4A CN201810130397A CN108419080B CN 108419080 B CN108419080 B CN 108419080B CN 201810130397 A CN201810130397 A CN 201810130397A CN 108419080 B CN108419080 B CN 108419080B
- Authority
- CN
- China
- Prior art keywords
- pixel
- values
- image
- context template
- context
- 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
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
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/182—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a pixel
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
- G06T9/001—Model-based coding, e.g. wire frame
-
- 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
- H04N19/42—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by implementation details or hardware specially adapted for video compression or decompression, e.g. dedicated software implementation
-
- 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
- H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive 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
- H04N19/70—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals characterised by syntax aspects related to video coding, e.g. related to compression standards
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
- H04N19/93—Run-length coding
Abstract
The invention relates to a method and a device for pipelined optimization of JPEGLS context calculation, in particular to a method and a device for evaluating a k-th pixel of an image and a related context template in a k-th clock cycle, sequentially reading values of the k-th pixel of the image and the related context template in cycles from a k +1 clock cycle to a k +5 clock cycle, and simultaneously evaluating the k +1 pixel of the image and the related context template until the reading of the value of the k +3 pixel and the evaluation of the k +4 pixel are completed, wherein k is a positive integer. When the context modeling is carried out, the feedback loops existing in the calculation in the template value are reduced, the calculation complexity of the context modeling is reduced, and the pipeline calculation performance of the encoder is effectively improved under the condition of little influence on the compression rate.
Description
Technical Field
The invention relates to a pipelining optimization method and device, in particular to a pipelining optimization method and device for JPEGLS context calculation.
Background
JPEG-LS is a lossless/near-lossless compression standard for continuous tone still images, ISO-14495-1/ITU-T.87. Its core algorithm is LOCO-I (Low Complexity loss Compression for images). The LOCO-I algorithm can obtain compression efficiency similar to or even better than that of a plurality of current compression algorithms based on arithmetic coding, can keep lower complexity at the same time, and is widely applied to the fields of digital cameras, network transmission, wireless communication, medical imaging and the like.
The JPEG-LS coding block diagram is shown in FIG. 1, and includes several steps of context modeling, prediction, normal mode coding, run-length mode coding, etc.
In the JPEG-LS standard, the coding model is based on the concept of a so-called "context". In context modeling, each sample value is conditioned on a small fraction of its neighbors. The modeling process is used to determine a probability distribution for encoding the current sample. Fig. 2 is a schematic diagram showing the relationship between a current sample x and its adjacent sample positions. Based on the reconstructed values of the samples at a, b, c and d, the context first decides whether the information of sample x is encoded in normal mode or run length mode.
Context modeling requires detection of boundaries in the vertical and horizontal directions. When the first row of the image data is coded, b, c and d do not exist, the three values are set to be 0, when the current data is positioned at the head of a column, a and c do not exist, when the data is positioned at the tail of the column, d does not exist, a and d obtain a reconstructed value b by taking b, and c obtains a reconstructed value a by taking a of the previous row. As shown in fig. 3, the boxes are image data with codes.
The values of the traditional methods a and d need to be obtained after the value b is obtained. In pipeline design, the context of the current pixel may lag one clock to acquire the complete template. The pipeline schematic is shown in FIG. 4:
when the first row of the image data is coded, b, c and d do not exist, the three values are set to be 0, when the current data is positioned at the head of a column, a and c do not exist, when the data is positioned at the tail of the column, d does not exist, a and d obtain a reconstructed value b by taking b, and c obtains a reconstructed value a by taking a of the previous row. As shown in fig. 3, the boxes are image data with codes.
Taking an image with a resolution of m × n as an example, Ix (1,1) represents pixel values of a first row and a first column of the image, and Ix (x, y) represents pixel values of an x-th row and a y-th column of the image. prevRb denotes the last row Rb.
When x is 1, Rb is 0; rc is 0; rd ═ 0;
when y is 1, Ra is prevRb;
when y is n, Rd is prevRb;
as can be seen from the above analysis, the conventional template value for context modeling needs to cache the previous line value of the current pixel of the image to obtain Rb, and when the boundary is determined, the Rc and Ra values need to obtain the reconstructed value according to the cached previous line value Rb.
Disclosure of Invention
The invention optimizes the algorithm aiming at the context modeling in the lossless/near lossless image compression coding in the JPEG-LS standard, reduces the feedback loop existing in the calculation in the template value during the context modeling, reduces the calculation complexity of the context modeling, and effectively improves the pipeline calculation performance of the encoder under the condition of little influence on the compression ratio. .
The technical problem of the invention is mainly solved by the following technical scheme:
a JPEGLS context calculation pipelining optimization method is characterized in that assignment is carried out on a kth pixel of an image and a related context template in a kth clock cycle, values of the kth pixel of the image and the related context template are sequentially read in cycles from a (k +1) th clock cycle to a (k + 5) th clock cycle, and assignment is carried out on a (k +1) th pixel of the image and the related context template at the same time until the reading of the value of the (k +3) th pixel and the assignment of the (k +4) th pixel are completed, wherein k is a positive integer.
In the foregoing pipelined optimization method for JPEGLS context calculation, in the k-th clock cycle, the k-th pixel of the image and its associated context template are assigned, i.e., Ix (k, k), Ra (k, k), Rb (k, k), Rc (k, k), Rd (k, k) are assigned; wherein the content of the first and second substances,
ix (x, y) represents the pixel value of the x-th row and the y-th column of the image;
ra (x, y) represents the reconstructed value of a in the context template of Ix (x, y);
rb (x, y) represents a reconstructed value of b in the context template of Ix (x, y);
rc (x, y) represents the reconstructed value of c in the context template of Ix (x, y);
rd (x, y) represents the reconstructed value of d in the context template for Ix (x, y);
when x is 1, Rb is 0; rc is 0; rd ═ 0;
when y is 1, Ra is 0;
when y is equal to n, Rd is equal to 0.
In the foregoing pipelined optimization method of JPEGLS context calculation, in the (k +1) th clock cycle, values of the k-th pixel of the image and its associated context template are read, that is, values of Ix (k, k), Ra (k, k), Rb (k, k), Rc (k, k), and Rd (k, k) are read, and values of the (k +1) th pixel of the image and its associated context template are assigned, that is, values of Ix (k, k +1), Ra (k, k +1), Rb (k, k +1), Rc (k, k +1), and Rd (k, k +1) are assigned.
In the foregoing method for pipelined optimization of JPEGLS context calculation, in the (k +2) th clock cycle, values of the (k +1) th pixel of the image and its associated context template are read, that is, values of Ix (k, k +1), Ra (k, k +1), Rb (k, k +1), Rc (k, k +1), and Rd (k, k +1) are read, and values of the (k +2) th pixel of the image and its associated context template are assigned, that is, values of Ix (k, k +2), Ra (k, k +2), Rb (k, k +2), Rc (k, k +2), and Rd (k, k +2) are assigned.
In the foregoing method for pipelined optimization of JPEGLS context calculation, in the (k +3) th clock cycle, values of the (k +2) th pixel of the image and its associated context template are read, that is, values of Ix (k, k +2), Ra (k, k +2), Rb (k, k +2), Rc (k, k +2), and Rd (k, k +2) are read, and values of the (k +3) th pixel of the image and its associated context template are assigned, that is, values of Ix (k, k +3), Ra (k, k +3), Rb (k, k +3), Rc (k, k +3), and Rd (k, k +3) are assigned.
In the foregoing method for pipelined optimization of JPEGLS context calculation, in the k +4 clock cycle, values of the k +3 th pixel of the image and the associated context template are read, that is, values of Ix (k, k +3), Ra (k, k +3), Rb (k, k +3), Rc (k, k +3), and Rd (k, k +3) are read, and values of the k +4 th pixel of the image and the associated context template are assigned, that is, values of Ix (k, k +4), Ra (k, k +4), Rb (k, k +4), Rc (k, k +4), and Rd (k, k +4) are assigned.
In the above-mentioned pipelined optimization method of JPEGLS context calculation, the kth pixel of the current image refers to the kth row and the kth column of the current image, … …, and the k +4 th pixel of the current image refers to the kth row and the kth +4 column of the current image, respectively.
An apparatus for pipelined optimization of JPEGLS context calculation, comprising:
a first assignment module: configured to assign a value to a kth pixel of the image and its associated context template at a kth clock cycle,
the numerical value reading and assigning module: : the method comprises the steps of sequentially reading the value of the kth pixel of an image and the related context template thereof in cycles from the (k +1) th clock cycle to the (k + 5) th clock cycle, and simultaneously assigning values to the (k +1) th pixel of the image and the related context template thereof until the reading of the value of the (k +3) th pixel and the assignment of the (k +4) th pixel are completed, wherein k is a positive integer.
In the foregoing pipelined optimization apparatus for JPEGLS context calculation, a specific operation process of the value reading and assigning module includes:
in the (k +1) th clock cycle, reading the values of the k-th pixel of the image and the related context template, namely reading the values of Ix (k, k), Ra (k, k), Rb (k, k), Rc (k, k), Rd (k, k), and assigning the k + 1-th pixel of the image and the related context template, namely assigning Ix (k, k +1), Ra (k, k +1), Rb (k, k +1), Rc (k, k +1), Rd (k, k + 1);
……
in the (k +4) th clock cycle, the values of the (k +3) th pixel of the image and its associated context template are read, i.e., the values of Ix (k, k +3), Ra (k, k +3), Rb (k, k +3), Rc (k, k +3), Rd (k, k +3) are read, and the values of the (k +4) th pixel of the image and its associated context template are assigned, i.e., the values of Ix (k, k +4), Ra (k, k +4), Rb (k, k +4), Rc (k, k +4), Rd (k, k +4) are assigned.
An apparatus for pipelined optimization of boundary processing for context calculation, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor is configured to perform the steps of the optimization method when executing the computer program.
Therefore, the invention has the following advantages: 1. the method simplifies the JPEGLS context modeling, and the context template is optimized during boundary processing. 2. When the context template is processed on the boundary, the dependency relationship among data in the template is eliminated, feedback loops existing in template values are reduced, and the computational complexity of context modeling is reduced. 3. After the optimization algorithm is implemented, the pipeline computing performance of the encoder is effectively improved under the condition that the influence on the image compression rate is small. .
Drawings
FIG. 1 is a JPEG-LS encoding flow chart.
FIG. 2 is a causal template for context modeling and prediction.
Fig. 3 is an example of image data boundary processing.
FIG. 4 is a schematic diagram of a template value pipeline of a conventional context modeling method.
Fig. 5 is a diagram illustrating an example of image data boundary processing in the present invention.
FIG. 6 is a schematic diagram of an improved context modeling method template value pipeline of the present invention.
Detailed Description
The technical scheme of the invention is further specifically described by the following embodiments and the accompanying drawings.
Example (b):
in order to realize the pipeline calculation, the invention can simplify the value of the context template during the boundary processing. During boundary processing, when data is located at the head or the tail of a column, values of a and d are 0, and value of c is still 0 because a of the previous row is also taken to obtain a reconstructed value a. The improved context template is shown in fig. 5, and the image data in the box:
taking an image with a resolution of m × n as an example, Ix (1,1) represents pixel values of a first row and a first column of the image, and Ix (x, y) represents pixel values of an x-th row and a y-th column of the image.
When x is 1, Rb is 0; rc is 0; rd ═ 0;
when y is 1, Ra is 0;
when y is equal to n, Rd is equal to 0;
the schematic diagram of the improved pipeline is shown in fig. 6:
the improved context template value taking method can realize complete pipeline calculation.
As described above, we take an image with a resolution of m × n as an example,
ix (1,1) represents the pixel value of the first row, first column of the image,
ra (1,1) represents the reconstructed value of a in the context template of Ix (1,1),
rb (1,1) represents the reconstructed value of b in the context template of Ix (1,1),
rc (1,1) represents the reconstructed value of c in the context template of Ix (1,1),
rd (1,1) represents the reconstructed value of d in the context template for Ix (1,1),
by the way of analogy, the method can be used,
ix (x, y) represents the x-th row and y-th column pixel values of the image,
ra (x, y) represents the reconstructed value of a in the context template of Ix (x, y),
rb (x, y) represents the reconstructed value of b in the context template of Ix (x, y),
rc (x, y) represents the reconstructed value of c in the context template of Ix (x, y),
rd (x, y) represents the reconstructed value of d in the context template for Ix (x, y),
the operation steps are as follows:
1) in the first clock cycle, the first pixel (first row, first column) of the image and its associated context template are assigned, i.e., Ix (1,1), Ra (1,1), Rb (1,1), Rc (1,1), Rd (1, 1).
2) In the second clock cycle, the values of the first pixel (first row, first column) of the image and its associated context template are read, i.e., the values of Ix (1,1), Ra (1,1), Rb (1,1), Rc (1,1), Rd (1,1) are read, and the values of the second pixel (first row, second column) of the image and its associated context template are assigned, i.e., the values of Ix (1,2), Ra (1,2), Rb (1,2), Rc (1,2), Rd (1,2) are assigned.
3) In the third clock cycle, the values of the second pixel (first row, second column) of the image and its associated context template are read, i.e., Ix (1,2), Ra (1,2), Rb (1,2), Rc (1,2), Rd (1,2) are read, and the values of the third pixel (first row, third column) of the image and its associated context template are assigned, i.e., Ix (1,3), Ra (1,3), Rb (1,3), Rc (1,3), Rd (1,3) are assigned.
4) In the fourth clock cycle, the values of the third pixel (first row, third column) of the image and its associated context template, i.e. Ix (1,3), Ra (1,3), Rb (1,3), Rc (1,3), Rd (1,3) are read, and the values of the fourth pixel (first row, fourth column) of the image and its associated context template, i.e. Ix (1,4), Ra (1,4), Rb (1,4), Rc (1,4), Rd (1,4) are assigned.
5) In the fifth clock cycle, the values of the fourth pixel (first row, fourth column) of the image and its associated context template, i.e. Ix (1,4), Ra (1,4), Rb (1,4), Rc (1,4), Rd (1,4) are read, and the values of the fifth pixel (first row, fifth column) of the image and its associated context template, i.e. Ix (1,5), Ra (1,5), Rb (1,5), Rc (1,5), Rd (1,5) are assigned.
By parity of reasoning, the writing and reading pipelining operation of the pixels of the whole image and the context template can be completed.
By comparing the improved compression effect of the context template value taking method, 5 single-channel 29M images are taken for testing, and the test results are shown in the following table:
single channel | Before improvement (byte) | After improvement (byte) | Change in |
1 | 4655827 | 4656164 | 0.007% |
2 | 9992242 | 9992479 | 0.002% |
3 | 11369745 | 11370224 | 0.004% |
4 | 11209168 | 11208816 | -0.003% |
5 | 12389960 | 12388711 | -0.010% |
As shown in the above table, the improved compression ratio is within ± 0.015%, and it can be seen that the improved contextual template value-taking method has little influence on the image compression ratio.
For decoding, the same context template value taking method can be adopted to realize lossless compression decoding.
The specific embodiments described herein are merely illustrative of the spirit of the invention. Various modifications or additions may be made to the described embodiments or alternatives may be employed by those skilled in the art without departing from the spirit or ambit of the invention as defined in the appended claims.
Claims (9)
1. A JPEGLS context calculation pipelining optimization method is characterized in that in the kth clock cycle, the k pixel of an image and a related context template are assigned, in the cycle from the k +1 clock cycle to the k +5 clock cycle, the values of the k pixel of the image and the related context template are sequentially read, and the values of the k +1 pixel of the image and the related context template are assigned at the same time until the reading of the value of the k +3 pixel and the assignment of the k +4 pixel are completed, wherein k is a positive integer;
in the k clock period, assigning values to the k pixel of the image and the related context template, i.e. assigning values to Ix (k, k), Ra (k, k), Rb (k, k), Rc (k, k), Rd (k, k); wherein the content of the first and second substances,
ix (x, y) represents the pixel value of the x-th row and the y-th column of the image;
ra (x, y) represents the reconstructed value of a in the context template of Ix (x, y);
rb (x, y) represents a reconstructed value of b in the context template of Ix (x, y);
rc (x, y) represents the reconstructed value of c in the context template of Ix (x, y);
rd (x, y) represents the reconstructed value of d in the context template for Ix (x, y);
when x = 1, Rb = 0; rc = 0; rd = 0;
when y = 1, Ra = 0;
rd = 0 when y = n.
2. The method of claim 1, wherein in the k +1 clock cycle, values of the k-th pixel of the image and its associated context template are read, i.e. values of Ix (k, k), Ra (k, k), Rb (k, k), Rc (k, k), and Rd (k, k) are read, and values of the k +1 pixel of the image and its associated context template are assigned, i.e. values of Ix (k, k +1), Ra (k, k +1), Rb (k, k +1), Rc (k, k +1), and Rd (k, k +1) are assigned.
3. The method of claim 1, wherein in the k +2 clock cycles, values of the (k +1) th pixel of the image and its associated context template are read, i.e. values of Ix (k, k +1), Ra (k, k +1), Rb (k, k +1), Rc (k, k +1), and Rd (k, k +1) are read, and values of the (k +2) th pixel of the image and its associated context template are assigned, i.e. values of Ix (k, k +2), Ra (k, k +2), Rb (k, k +2), Rc (k, k +2), and Rd (k, k +2) are assigned.
4. The method of claim 1, wherein in the k +3 clock cycles, values of the k +2 th pixel of the image and its associated context template are read, i.e. values of Ix (k, k +2), Ra (k, k +2), Rb (k, k +2), Rc (k, k +2), and Rd (k, k +2) are read, and values of the k +3 th pixel of the image and its associated context template are assigned, i.e. values of Ix (k, k +3), Ra (k, k +3), Rb (k, k +3), Rc (k, k +3), and Rd (k, k +3) are assigned.
5. The method of claim 1, wherein in the k +4 clock cycles, values of k +3 pixels of the image and their associated context templates are read, i.e. values of Ix (k, k +3), Ra (k, k +3), Rb (k, k +3), Rc (k, k +3), and Rd (k, k +3) are read, and values of k +4 pixels of the image and their associated context templates are assigned, i.e. values of Ix (k, k +4), Ra (k, k +4), Rb (k, k +4), Rc (k, k +4), and Rd (k, k +4) are assigned.
6. The method of claim 1 in which the kth pixel of the current image is the kth row and kth column of the current image, … …, and the (k +4) th pixel of the current image is the kth row and the (k +4) th column of the current image.
7. An apparatus for pipelined optimization of JPEGLS context calculation, comprising:
a first assignment module: configured to assign a value to a kth pixel of the image and its associated context template at a kth clock cycle,
the numerical value reading and assigning module: the method comprises the steps that the method is configured to sequentially read the value of the kth pixel of an image and the value of a context template related to the kth pixel in cycles from the (k +1) th clock cycle to the (k + 5) th clock cycle, and assign values to the (k +1) th pixel of the image and the context template related to the kth pixel of the image at the same time until the reading of the value of the (k +3) th pixel and the assignment of the (k +4) th pixel are completed, wherein k is a positive integer;
in the k clock period, assigning values to the k pixel of the image and the related context template, i.e. assigning values to Ix (k, k), Ra (k, k), Rb (k, k), Rc (k, k), Rd (k, k); wherein the content of the first and second substances,
ix (x, y) represents the pixel value of the x-th row and the y-th column of the image;
ra (x, y) represents the reconstructed value of a in the context template of Ix (x, y);
rb (x, y) represents a reconstructed value of b in the context template of Ix (x, y);
rc (x, y) represents the reconstructed value of c in the context template of Ix (x, y);
rd (x, y) represents the reconstructed value of d in the context template for Ix (x, y);
when x = 1, Rb = 0; rc = 0; rd = 0;
when y = 1, Ra = 0;
rd = 0 when y = n.
8. The apparatus of claim 7, wherein the numerical reading and assigning module comprises:
in the (k +1) th clock cycle, reading the values of the k-th pixel of the image and the related context template, namely reading the values of Ix (k, k), Ra (k, k), Rb (k, k), Rc (k, k), Rd (k, k), and assigning the k + 1-th pixel of the image and the related context template, namely assigning Ix (k, k +1), Ra (k, k +1), Rb (k, k +1), Rc (k, k +1), Rd (k, k + 1);
……
in the (k +4) th clock cycle, the values of the (k +3) th pixel of the image and its associated context template are read, i.e., the values of Ix (k, k +3), Ra (k, k +3), Rb (k, k +3), Rc (k, k +3), Rd (k, k +3) are read, and the values of the (k +4) th pixel of the image and its associated context template are assigned, i.e., the values of Ix (k, k +4), Ra (k, k +4), Rb (k, k +4), Rc (k, k +4), Rd (k, k +4) are assigned.
9. An apparatus for pipelined optimization of boundary processing for context calculation, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor is configured to perform the steps of the optimization method according to any one of claims 1 to 6 when executing the computer program.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810130397.4A CN108419080B (en) | 2018-02-08 | 2018-02-08 | Method and device for streamline optimization of JPEGLS context calculation |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810130397.4A CN108419080B (en) | 2018-02-08 | 2018-02-08 | Method and device for streamline optimization of JPEGLS context calculation |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108419080A CN108419080A (en) | 2018-08-17 |
CN108419080B true CN108419080B (en) | 2020-10-13 |
Family
ID=63127060
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810130397.4A Active CN108419080B (en) | 2018-02-08 | 2018-02-08 | Method and device for streamline optimization of JPEGLS context calculation |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108419080B (en) |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103581666A (en) * | 2013-09-30 | 2014-02-12 | 西安空间无线电技术研究所 | Image compression run-length coding method for updating mask words successively |
CN105828070A (en) * | 2016-03-23 | 2016-08-03 | 华中科技大学 | Anti-error code propagation JPEG-LS image lossless/near-lossless compression algorithm hardware realization method |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP3990639B2 (en) * | 2003-01-24 | 2007-10-17 | 三菱電機株式会社 | Image processing apparatus, image processing method, and image display apparatus |
JP4587175B2 (en) * | 2005-05-19 | 2010-11-24 | キヤノン株式会社 | Image encoding apparatus and method, computer program, and computer-readable storage medium |
CN101534373B (en) * | 2009-04-24 | 2011-02-09 | 北京空间机电研究所 | Remote sensing image near-lossless compression hardware realization method based on improved JPEG-LS algorithm |
CN101771874B (en) * | 2009-12-31 | 2011-09-14 | 华中科技大学 | Satellite image compression method and device for realizing satellite image compression |
CN102970531B (en) * | 2012-10-19 | 2015-04-22 | 西安电子科技大学 | Method for implementing near-lossless image compression encoder hardware based on joint photographic experts group lossless and near-lossless compression of continuous-tone still image (JPEG-LS) |
-
2018
- 2018-02-08 CN CN201810130397.4A patent/CN108419080B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103581666A (en) * | 2013-09-30 | 2014-02-12 | 西安空间无线电技术研究所 | Image compression run-length coding method for updating mask words successively |
CN105828070A (en) * | 2016-03-23 | 2016-08-03 | 华中科技大学 | Anti-error code propagation JPEG-LS image lossless/near-lossless compression algorithm hardware realization method |
Also Published As
Publication number | Publication date |
---|---|
CN108419080A (en) | 2018-08-17 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US20190081637A1 (en) | Data inspection for compression/decompression configuration and data type determination | |
CN111797976B (en) | Training method, image recognition method, device, equipment and medium for neural network | |
US8615138B2 (en) | Image compression using sub-resolution images | |
CN109379598B (en) | Image lossless compression method based on FPGA | |
US20080298472A1 (en) | Throughput Performance When Applying Deblocking Filters On Reconstructed Image Frames | |
US11062210B2 (en) | Method and apparatus for training a neural network used for denoising | |
CN111641826B (en) | Method, device and system for encoding and decoding data | |
US20220222541A1 (en) | Neural Network Representation Formats | |
EP3841528A1 (en) | Data compression using integer neural networks | |
CN113874916A (en) | AI-assisted programmable hardware video codec | |
CN116917926A (en) | Indication of feature map data | |
CN110648348B (en) | Quantum image segmentation method based on NEQR expression | |
CN110913225B (en) | Image encoding method, image encoding device, electronic device, and computer-readable storage medium | |
CN112399177B (en) | Video coding method, device, computer equipment and storage medium | |
CN108419080B (en) | Method and device for streamline optimization of JPEGLS context calculation | |
CN114501031B (en) | Compression coding and decompression method and device | |
CN109474826B (en) | Picture compression method and device, electronic equipment and storage medium | |
KR100739110B1 (en) | Method for compressing binary image rapidly | |
CN115086660B (en) | Decoding and encoding method based on point cloud attribute prediction, decoder and encoder | |
Jubran et al. | Sequence-level reference frames in video coding | |
US10283073B2 (en) | Data processing systems | |
US20240046527A1 (en) | End-to-end optimization of adaptive spatial resampling towards machine vision | |
CN110189272B (en) | Method, apparatus, device and storage medium for processing image | |
WO2024001433A1 (en) | Encoding mode prediction method and apparatus, electronic device and storage medium | |
CN116668702B (en) | Video coding method, device, terminal equipment and storage medium |
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 |