CN108712656A - A kind of the long-distance video processing method and video service terminal of dynamic recognition image procossing - Google Patents
A kind of the long-distance video processing method and video service terminal of dynamic recognition image procossing Download PDFInfo
- Publication number
- CN108712656A CN108712656A CN201810581317.7A CN201810581317A CN108712656A CN 108712656 A CN108712656 A CN 108712656A CN 201810581317 A CN201810581317 A CN 201810581317A CN 108712656 A CN108712656 A CN 108712656A
- Authority
- CN
- China
- Prior art keywords
- grades
- thread
- dynamic recognition
- continuous
- flow type
- 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.)
- Granted
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
- H04N19/625—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding using discrete cosine transform [DCT]
-
- 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
- H04N19/436—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 using parallelised computational arrangements
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Computing Systems (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Discrete Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Image Processing (AREA)
- Image Analysis (AREA)
Abstract
The invention belongs to image dynamic recognition technical fields,More particularly to a kind of long-distance video processing method of dynamic recognition image procossing,Present invention simultaneously provides a kind of long-distance video service terminals of dynamic recognition image procossing,The data inputted parallel are inputted N grades of operation threads by the continuous-flow type concurrent operation method respectively,The corresponding image processing function of every level-one of the N grades of operation thread exports processing result image respectively,The accumulated add operation result for exporting image processing function multiplication result or add operation result or multi-stage cascade of the processing result image exported respectively,The present invention solves the problems, such as that the prior art exists since the serial poor efficiency for executing code of general processor is not suitable for the scientific calculation of image procossing,With the scientific calculation for being more suitable for image procossing,Improve the arithmetic speed of multiplier,Improve the execution efficiency of whole system,Practicability and the strong advantageous effects of application.
Description
Technical field
The invention belongs to the long-range of image dynamic recognition technical field more particularly to a kind of dynamic recognition image procossing
Method for processing video frequency, present invention simultaneously provides a kind of long-distance video service terminals of dynamic recognition image procossing.
Background technology
Currently, discrete cosine transform (DCT) is a kind of real number field transformation, transformation kernel is real number cosine function, to a pair
Many important visual informations in relation to image all concentrate on the sub-fraction coefficient of dct transform after image progress discrete cosine transform
In, therefore discrete cosine transform is the core of Image Lossy Compression JPEG, while the master of so-called " transform domain information hidden algorithm "
One of " transform domain " is wanted, because image procossing uses one-dimensional discrete cosine transform, the prior art to exist due to general processor
Not the problem of serial poor efficiency for executing code is not suitable for the scientific calculation of image procossing.
Invention content
The present invention provides a kind of the long-distance video processing method and video service terminal of dynamic recognition image procossing, with solution
The prior art is proposed in certainly above-mentioned background technology to exist since the serial poor efficiency for executing code of general processor is not applicable
In the scientific calculation of image procossing the problem of.
Technical problem solved by the invention is realized using following technical scheme：A kind of dynamic recognition image procossing
Long-distance video processing method, including：Continuous-flow type concurrent operation method, the continuous-flow type concurrent operation method include：
Continuous-flow type concurrent operation equation：
The n is the series set that parallel dynamic recognition is classified operation；
The N is the maximum series increasing one that parallel dynamic recognition is classified operation；
The operational equation of the x (k) is：
The operational equation of the e (n) is：
The F (n) is image processing function.
Further, the series collection of the classification operation is combined into 0~N-1 natural number set.
Further, the data inputted parallel are inputted N grades of operation threads by the continuous-flow type concurrent operation method respectively, described
The corresponding image processing function of every level-one of N grades of operation threads exports processing result image, the image exported respectively respectively
The add operation of the accumulated output image processing function multiplication result of handling result or add operation result or multi-stage cascade
As a result.
Further, the continuous-flow type concurrent operation method includes eight grades of concurrent operations of N=8.
Further, eight grades of concurrent operations include eight grades of operation threads.
Further, the add operation equation of the multi-stage cascade of eight grades of operation threads is：
F (8)=f (0)+f (1)+f (2)+f (3)+f (4)+f (5)+f (6)+f (7)；
The f (0) is first order operation thread；
The f (1) is second level operation thread；
The f (2) is third level operation thread；
The f (3) is fourth stage operation thread；
The f (4) is level V operation thread；
The f (5) is the 6th grade of operation thread；
The f (6) is the 7th grade of operation thread；
The f (7) is the 8th grade of operation thread；
Meanwhile the present invention also provides a kind of long-distance video service terminals of dynamic recognition image procossing, including continuous-flow type
Concurrent operation module, the continuous-flow type concurrent operation module include：
Continuous-flow type concurrent operation equation：
The n is the series set that parallel dynamic recognition is classified operation；
The N is the maximum series increasing one that parallel dynamic recognition is classified operation；
The operational equation of the x (k) is：
The operational equation of the e (n) is：
The F (n) is image processing function.
Further, the series collection of the classification operation is combined into 0~N-1 natural number set.
Further, the data inputted parallel are inputted N grades of operation threads by the continuous-flow type concurrent operation method respectively, described
The corresponding image processing function of every level-one of N grades of operation threads exports processing result image, the image exported respectively respectively
The add operation of the accumulated output image processing function multiplication result of handling result or add operation result or multi-stage cascade
As a result.
Further, the continuous-flow type concurrent operation method includes eight grades of concurrent operations of N=8, eight grades of concurrent operation packets
Eight grades of operation threads are included, the add operation equation of the multi-stage cascade of eight grades of operation threads is：
F (8)=f (0)+f (1)+f (2)+f (3)+f (4)+f (5)+f (6)+f (7)；
The f (0) is first order operation thread；
The f (1) is second level operation thread；
The f (2) is third level operation thread；
The f (3) is fourth stage operation thread；
The f (4) is level V operation thread；
The f (5) is the 6th grade of operation thread；
The f (6) is the 7th grade of operation thread；
The f (7) is the 8th grade of operation thread.
Advantageous effects：
1, this patent includes using the continuous-flow type concurrent operation method：
Continuous-flow type concurrent operation equation：
The n is the series set that parallel dynamic recognition is classified operation；
The N is the maximum series increasing one that parallel dynamic recognition is classified operation；
The operational equation of the x (k) is：
The operational equation of the e (n) is：
The F (n) is image processing function, and the continuous-flow type concurrent operation method inputs the data inputted parallel respectively
The corresponding image processing function of every level-one of N grades of operation threads, the N grades of operation thread exports processing result image respectively, institute
State the accumulated output image processing function multiplication result of the processing result image exported respectively or add operation result or more
The cascade add operation of grade realizes a major class figure as a result, since this algorithm has the data flow of the concurrency and rule of height
The Processing Algorithm of picture, the interaction that the algorithm realization based on systolic arrays at this time is actually converted to an operation node is realized, right
In may all nodes realized in one single chip, for algorithm complexity, the larger algorithm of node size can be with multiple threads pair
An operation node is answered, since the concurrency and flowing water type, many algorithms of pulsation processing can access real-time realization, because
This, the serial efficiency for executing code for avoiding general processor is low, is more suitable for the scientific calculation of image procossing.
2, this patent uses the fast algorithm of one-dimensional DCT, it can be seen that, entire operation uses parallel form from algorithm
8 input datas are handled simultaneously, and the systematicness of the data flow in structure and hierarchy show that flowing water can be used in algorithm
Mode handled, therefore, software operation can obtain higher operation efficiency, and the node in flow graph includes multiplication, addition
And delay cell, by realizing that multiplier and adder and latch can realize node operation in single-chip or program, soft
In part, adder unit is provided, can not only realize that the add operation in node unit, multiplying can also lead to using the unit
It crosses multistage adder stage connection to realize, improves the arithmetic speed of multiplier.
3, this patent can obtain efficient realization, still, one for some specific algorithm using suitable hardware configuration
A Scientific Calculation Program includes many places intensive operations code, and each section of code is accomplished that different algorithms, in order to mitigate general place
It manages the burden of device and improves operation efficiency, hardware or software realization can be used to each section of intensive operations code, this is in journey
The operation phase of sequence constantly changes circuit structure or changes the configuration of thread, this is also to reconfigure mode, the transfer of this partial code
It is executed on to rational hardware configuration, improves the execution efficiency of whole system.
4, this patent realizes continuous-flow type concurrent operation method using continuous-flow type concurrent operation module, and this improves dynamics
Reconfigure the practicability and application of image processing apparatus.
Description of the drawings
Fig. 1 is a kind of flow chart of the long-distance video processing method of dynamic recognition image procossing of the present invention.
Specific implementation mode
The present invention is described further below in conjunction with attached drawing：
In figure：
The data inputted parallel are inputted N grades of operation threads by S101- continuous-flow type concurrent operation methods respectively；
The corresponding image processing function of every level-one of N described in S102- grades of operation thread exports processing result image respectively；
The accumulated output image processing function multiplication result of processing result image that is exported respectively described in S103- adds
The add operation result of method operation result or multi-stage cascade；
Embodiment：
The present embodiment：As shown in Figure 1, a kind of 1, long-distance video processing method of dynamic recognition image procossing, feature
It is, including：Continuous-flow type concurrent operation method, the continuous-flow type concurrent operation method include：
Continuous-flow type concurrent operation equation：
The n is the series set that parallel dynamic recognition is classified operation；
The N is the maximum series increasing one that parallel dynamic recognition is classified operation；
The operational equation of the x (k) is：
The operational equation of the e (n) is：
The F (n) is image processing function.
The series collection of the classification operation is combined into 0~N-1 natural number set.
The data inputted parallel are inputted N grades of operation thread S101 by the continuous-flow type concurrent operation method respectively, N grades described
The corresponding image processing function of every level-one of operation thread exports processing result image S102, the image exported respectively respectively
The add operation of the accumulated output image processing function multiplication result of handling result or add operation result or multi-stage cascade
As a result S103.
Due to including using the continuous-flow type concurrent operation method：
Continuous-flow type concurrent operation equation：
The n is the series set that parallel dynamic recognition is classified operation；
The N is the maximum series increasing one that parallel dynamic recognition is classified operation；
The operational equation of the x (k) is：
The operational equation of the e (n) is：
The F (n) is image processing function, and the continuous-flow type concurrent operation method inputs the data inputted parallel respectively
The corresponding image processing function of every level-one of N grades of operation threads, the N grades of operation thread exports processing result image respectively, institute
State the accumulated output image processing function multiplication result of the processing result image exported respectively or add operation result or more
The cascade add operation of grade realizes a major class figure as a result, since this algorithm has the data flow of the concurrency and rule of height
The Processing Algorithm of picture, the interaction that the algorithm realization based on systolic arrays at this time is actually converted to an operation node is realized, right
In may all nodes realized in one single chip, for algorithm complexity, the larger algorithm of node size can be with multiple threads pair
An operation node is answered, since the concurrency and flowing water type, many algorithms of pulsation processing can access real-time realization, because
This, the serial efficiency for executing code for avoiding general processor is low, is more suitable for the scientific calculation of image procossing.
The continuous-flow type concurrent operation method includes eight grades of concurrent operations of N=8.
Due to the fast algorithm using one-dimensional DCT, it can be seen that, entire operation uses parallel form pair 8 from algorithm
A input data is handled simultaneously, and the systematicness of the data flow in structure and hierarchy show that the side of flowing water can be used in algorithm
Formula is handled, and therefore, software operation can obtain higher operation efficiency, and the node in flow graph includes multiplication, addition and prolongs
Slow unit, by realizing that multiplier and adder and latch can realize node operation in single-chip or program, in software,
Adder unit is provided, can not only realize that add operation, the multiplying in node unit can also be by more using the unit
Grade adder cascade is realized, the arithmetic speed of multiplier is improved.
Eight grades of concurrent operations include eight grades of operation threads.
Since for some specific algorithm, efficient realization can be obtained using suitable hardware configuration, still, a section
It includes many places intensive operations code that program is calculated in student movement, and each section of code is accomplished that different algorithms, in order to mitigate general processor
Burden and improve operation efficiency, to each section of intensive operations code can use hardware or software realization, this is in program
Operation phase constantly changes circuit structure or changes the configuration of thread, this is also to reconfigure mode, this partial code is transferred to conjunction
It is executed on the hardware configuration of reason, improves the execution efficiency of whole system.
The add operation equation of the multi-stage cascade of eight grades of operation threads is：
F (8)=f (0)+f (1)+f (2)+f (3)+f (4)+f (5)+f (6)+f (7)；
The f (0) is first order operation thread；
The f (1) is second level operation thread；
The f (2) is third level operation thread；
The f (3) is fourth stage operation thread；
The f (4) is level V operation thread；
The f (5) is the 6th grade of operation thread；
The f (6) is the 7th grade of operation thread；
The f (7) is the 8th grade of operation thread；
Including continuous-flow type concurrent operation module, the continuous-flow type concurrent operation module includes：
Continuous-flow type concurrent operation equation：
The n is parallel dynamic recognition
The N is the maximum series increasing one that parallel dynamic recognition is classified operation；
The operational equation of the x (k) is：
The operational equation of the e (n) is：
The F (n) is image processing function；
Due to realizing continuous-flow type concurrent operation method using continuous-flow type concurrent operation module, this improves dynamics to re-match
Set the practicability and application of image processing apparatus.
The series collection of the classification operation is combined into 0~N-1 natural number set.
The data inputted parallel are inputted N grades of operation threads, the N grades of operation by the continuous-flow type concurrent operation method respectively
The corresponding image processing function of every level-one of thread exports processing result image, the processing result image exported respectively respectively
The add operation result of accumulated output image processing function multiplication result or add operation result or multi-stage cascade.
The continuous-flow type concurrent operation method includes eight grades of concurrent operations of N=8, and eight grades of concurrent operations include eight grades of fortune
Thread is calculated, the add operation equation of the multi-stage cascade of eight grades of operation threads is：
F (8)=f (0)+f (1)+f (2)+f (3)+f (4)+f (5)+f (6)+f (7)；
The f (0) is first order operation thread；
The f (1) is second level operation thread；
The f (2) is third level operation thread；
The f (3) is fourth stage operation thread；
The f (4) is level V operation thread；
The f (5) is the 6th grade of operation thread；
The f (6) is the 7th grade of operation thread；
The f (7) is the 8th grade of operation thread.
Operation principle：
This patent includes by the continuous-flow type concurrent operation method：
Continuous-flow type concurrent operation equation：
The n is the series set that parallel dynamic recognition is classified operation；
The N is the maximum series increasing one that parallel dynamic recognition is classified operation；
The operational equation of the x (k) is：
The operational equation of the e (n) is：
The F (n) is image processing function, and the continuous-flow type concurrent operation method inputs the data inputted parallel respectively
The corresponding image processing function of every level-one of N grades of operation threads, the N grades of operation thread exports processing result image respectively, institute
State the accumulated output image processing function multiplication result of the processing result image exported respectively or add operation result or more
The cascade add operation of grade realizes a major class figure as a result, since this algorithm has the data flow of the concurrency and rule of height
The Processing Algorithm of picture, the interaction that the algorithm realization based on systolic arrays at this time is actually converted to an operation node is realized, right
In may all nodes realized in one single chip, for algorithm complexity, the larger algorithm of node size can be with multiple threads pair
An operation node is answered, since the concurrency and flowing water type, many algorithms of pulsation processing can access real-time realization, this hair
The bright poor efficiency for solving the prior art there are the prior art in the presence of the serial execution code due to general processor is not suitable for
The problem of scientific calculation of image procossing, has the operation speed for being more suitable for the scientific calculation of image procossing, improving multiplier
Spend, improve the execution efficiency of whole system, the advantageous effects of practicability and application.
Using technical scheme of the present invention or those skilled in the art under the inspiration of technical solution of the present invention, design
Go out similar technical solution, and reach above-mentioned technique effect, is to fall into protection scope of the present invention.
Claims (10)
1. a kind of long-distance video processing method of dynamic recognition image procossing, which is characterized in that including：Continuous-flow type concurrent operation
Method, the continuous-flow type concurrent operation method include：
Continuous-flow type concurrent operation equation：
The n is the series set that parallel dynamic recognition is classified operation；
The N is the maximum series increasing one that parallel dynamic recognition is classified operation；
The operational equation of the x (k) is：
The operational equation of the e (n) is：
The F (n) is image processing function.
2. a kind of long-distance video processing method of dynamic recognition image procossing according to claim 1, which is characterized in that
The series collection of the classification operation is combined into 0~N-1 natural number set.
3. a kind of long-distance video processing method of dynamic recognition image procossing according to claim 2, which is characterized in that
The data inputted parallel are inputted N grades of operation threads by the continuous-flow type concurrent operation method respectively, the N grades of operation thread it is every
The corresponding image processing function of level-one exports processing result image respectively, and the processing result image exported respectively is accumulated defeated
Go out image processing function multiplication result or add operation result or the add operation result of multi-stage cascade.
4. a kind of long-distance video processing method of dynamic recognition image procossing according to claim 3, which is characterized in that
The continuous-flow type concurrent operation method includes eight grades of concurrent operations of N=8.
5. a kind of long-distance video processing method of dynamic recognition image procossing according to claim 4, which is characterized in that
Eight grades of concurrent operations include eight grades of operation threads.
6. a kind of long-distance video processing method of dynamic recognition image procossing according to claim 5, which is characterized in that
The add operation equation of the multi-stage cascade of eight grades of operation threads is：
F (8)=f (0)+f (1)+f (2)+f (3)+f (4)+f (5)+f (6)+f (7)；
The f (0) is first order operation thread；
The f (1) is second level operation thread；
The f (2) is third level operation thread；
The f (3) is fourth stage operation thread；
The f (4) is level V operation thread；
The f (5) is the 6th grade of operation thread；
The f (6) is the 7th grade of operation thread；
The f (7) is the 8th grade of operation thread.
7. a kind of long-distance video service terminal of dynamic recognition image procossing, which is characterized in that including continuous-flow type concurrent operation
Module, the continuous-flow type concurrent operation module include：
Continuous-flow type concurrent operation equation：
The n is the series set that parallel dynamic recognition is classified operation；
The N is the maximum series increasing one that parallel dynamic recognition is classified operation；
The operational equation of the x (k) is：
The operational equation of the e (n) is：
The F (n) is image processing function.
8. a kind of long-distance video service terminal of dynamic recognition image procossing according to claim 1, which is characterized in that
The series collection of the classification operation is combined into 0~N-1 natural number set.
9. a kind of long-distance video service terminal of dynamic recognition image procossing according to claim 1, which is characterized in that
The data inputted parallel are inputted N grades of operation threads by the continuous-flow type concurrent operation method respectively, the N grades of operation thread it is every
The corresponding image processing function of level-one exports processing result image respectively, and the processing result image exported respectively is accumulated defeated
Go out image processing function multiplication result or add operation result or the add operation result of multi-stage cascade.
10. a kind of long-distance video service terminal of dynamic recognition image procossing according to claim 1, feature exist
In the continuous-flow type concurrent operation method includes eight grades of concurrent operations of N=8, and eight grades of concurrent operations include eight grades of operation lines
The add operation equation of journey, the multi-stage cascade of eight grades of operation threads is：
F (8)=f (0)+f (1)+f (2)+f (3)+f (4)+f (5)+f (6)+f (7)；
The f (0) is first order operation thread；
The f (1) is second level operation thread；
The f (2) is third level operation thread；
The f (3) is fourth stage operation thread；
The f (4) is level V operation thread；
The f (5) is the 6th grade of operation thread；
The f (6) is the 7th grade of operation thread；
The f (7) is the 8th grade of operation thread.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810581317.7A CN108712656B (en) | 2018-06-07 | 2018-06-07 | Remote video processing method and video service terminal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810581317.7A CN108712656B (en) | 2018-06-07 | 2018-06-07 | Remote video processing method and video service terminal |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108712656A true CN108712656A (en) | 2018-10-26 |
| CN108712656B CN108712656B (en) | 2021-07-20 |
Family
ID=63871464
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810581317.7A Active CN108712656B (en) | 2018-06-07 | 2018-06-07 | Remote video processing method and video service terminal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108712656B (en) |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101227607A (en) * | 2007-12-12 | 2008-07-23 | 深圳市同洲电子股份有限公司 | Exhaustive motion estimation searching method and apparatus based on parallel pipeline |
| US20090040946A1 (en) * | 2007-08-06 | 2009-02-12 | Archer Charles J | Executing an Allgather Operation on a Parallel Computer |
| CN102007768A (en) * | 2008-04-16 | 2011-04-06 | 弗朗霍夫应用科学研究促进协会 | Bit-depth scalability |
| US20170255515A1 (en) * | 2016-03-04 | 2017-09-07 | Samsung Electronics Co., Ltd. | Method and apparatus for performing data recovery in a raid storage |
| CN107562449A (en) * | 2016-06-30 | 2018-01-09 | 欧姆龙株式会社 | Image processing apparatus, image processing method and image processing program |
-
2018
- 2018-06-07 CN CN201810581317.7A patent/CN108712656B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20090040946A1 (en) * | 2007-08-06 | 2009-02-12 | Archer Charles J | Executing an Allgather Operation on a Parallel Computer |
| CN101227607A (en) * | 2007-12-12 | 2008-07-23 | 深圳市同洲电子股份有限公司 | Exhaustive motion estimation searching method and apparatus based on parallel pipeline |
| CN102007768A (en) * | 2008-04-16 | 2011-04-06 | 弗朗霍夫应用科学研究促进协会 | Bit-depth scalability |
| US20170255515A1 (en) * | 2016-03-04 | 2017-09-07 | Samsung Electronics Co., Ltd. | Method and apparatus for performing data recovery in a raid storage |
| CN107562449A (en) * | 2016-06-30 | 2018-01-09 | 欧姆龙株式会社 | Image processing apparatus, image processing method and image processing program |
Non-Patent Citations (2)
| Title |
|---|
| HU YONG-SHENG: "Realization of GA-NN blind equalization algorithm", 《IEEE》 * |
| 寇福东: "基于多核DSP的数字图像处理并行化方法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108712656B (en) | 2021-07-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | Sparse-YOLO: Hardware/software co-design of an FPGA accelerator for YOLOv2 | |
| US11531540B2 (en) | Processing apparatus and processing method with dynamically configurable operation bit width | |
| EP3443465A2 (en) | Method, system and program product for mask-based compression of a sparse matrix | |
| CN109117187A (en) | Convolutional neural networks accelerated method and relevant device | |
| CN106095588A (en) | CDVS based on GPGPU platform extracts process accelerated method | |
| CN109409514A (en) | Fixed-point calculation method, apparatus, equipment and the storage medium of convolutional neural networks | |
| JPH11203271A (en) | Dct circuit, idct circuit and dct/idct circuit | |
| CN113033794B (en) | Light weight neural network hardware accelerator based on deep separable convolution | |
| Yan et al. | An efficient sparse-dense matrix multiplication on a multicore system | |
| CN112181657A (en) | Video processing method and device, electronic equipment and storage medium | |
| Ujiie et al. | Approximated prediction strategy for reducing power consumption of convolutional neural network processor | |
| Xiao et al. | FPGA-based scalable and highly concurrent convolutional neural network acceleration | |
| Ju et al. | Threshnet: An efficient densenet using threshold mechanism to reduce connections | |
| CN102567282A (en) | Device and method for realizing FFT (Fast Fourier Transform) calculation in universal DSP (Digital Signal Processor) | |
| CN108712656A (en) | A kind of the long-distance video processing method and video service terminal of dynamic recognition image procossing | |
| CN102970545A (en) | Static image compression method based on two-dimensional discrete wavelet transform algorithm | |
| CN109992742A (en) | A kind of signal processing method and device | |
| CN109740729B (en) | Operation method, device and related product | |
| CN110716751B (en) | High-parallelism computing platform, system and computing implementation method | |
| CN109711538B (en) | Operation method, device and related product | |
| Jiang et al. | A distributed dynamic parallel algorithm for SIFT feature extraction | |
| CN104901651A (en) | Realizing circuit and method of digital filter | |
| Xu et al. | Efficient hardware implementation of cellular neural networks with powers-of-two based incremental quantization | |
| Messom et al. | Stream processing of integral images for real-time object detection | |
| Thepade et al. | Performance augmentation of video retrieval using even-odd videos with multilevel block truncation coding |
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 |
Effective date of registration: 20221128 Address after: 250013 Room 903, Block A, Jinyu Center, No. 5777, Jingshi Road, Jinan District, China (Shandong) Pilot Free Trade Zone, Jinan City, Shandong Province Patentee after: Shandong Qingcheng Shilian Information Technology Co.,Ltd. Address before: 256603, No. five, 391 the Yellow River Road, Bincheng District, Shandong, Binzhou Patentee before: BINZHOU University |
|
| TR01 | Transfer of patent right |