CN105357523B - One kind being based on HOSVD algorithm video compression system and method - Google Patents
One kind being based on HOSVD algorithm video compression system and method Download PDFInfo
- Publication number
- CN105357523B CN105357523B CN201510677893.8A CN201510677893A CN105357523B CN 105357523 B CN105357523 B CN 105357523B CN 201510677893 A CN201510677893 A CN 201510677893A CN 105357523 B CN105357523 B CN 105357523B
- Authority
- CN
- China
- Prior art keywords
- video
- hosvd
- frame
- compression system
- processor
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired - Fee Related
Links
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The present invention is a kind of based on HOSVD algorithm video compression system and method, the video compression system includes: camera device and respective sensor, the camera device and respective sensor are connected separately with CPLD processor and FIFO memory, the CPLD processor connects DSP digital signal processor with FIFO memory, the CPLD processor is connected with FIFO memory and AVS encoder respectively, and the AVS encoder is connected with ROM video memory.The present invention uses tensor resolution theory software and hardware combining, realizes time domain and spatial redundancy Information Compression, maximum compression data-storing amount, to realize the effective preservation and minimum information loss of color video.
Description
Technical field
The present invention relates to monitored video compression technical fields, and in particular to one kind is based on HOSVD algorithm video compression system
And method.
Background technique
In some public places such as residential quarter, bank, airport waiting room, shopping center, traffic intersection, office building, campus
With residential quarter etc., monitoring device is very universal, and this aspect is that the lives and properties of people and public safety provide guarantee, together
When be also that relevant unit increases administrative burden and financial burden: daily 24 hours, annual more than 360 days monitor video fluxions
According to, to monitoring unit increase the costs such as a large amount of labor management, plant maintenance, data-storing and data processing, be unit band
Carry out huge financial burden.Some units are to cut down expenses, and deliberately close monitoring device, or monitoring device is allowed to locate idle shape for a long time
State, monitoring device perform practically no function, and leave many security risks.Currently, monitoring in the market include closed-circuit control, video monitoring,
The monitoring systems such as video recording and picture control, long-range monitoring, infrared monitoring and network monitoring.Under normal circumstances, every monitoring device
86400 hours of continuous work are needed within 1 year, by high definition 720P(1280 × 720) for format single channel video, storage 1 in 24 hours
Its video data volume about 42G, (calculating by 365 days) data volume is about 15TB (15330GB) within 1 year, and it is empty to occupy huge storage
Between, height is required to storage facilities, processing and lookup data take time and effort, and data save difficult.But after such as being carried out to monitoring data
Phase excess compression, then video pictures quality can be badly damaged, and substantially reduce the effective percentage of video monitoring.On the other hand, monitoring system
The a large amount of lengthy and jumbled information obtained of uniting generate interference to main information, increase the difficulty of data analysis, measure the information that need to be obtained
Now excessive redundancy brings challenge to information preservation and data analysis.
With the most similar implementation of the present invention, i.e. AVS compress technique [1].
AVS core technology is by entropy decoding, reorders, inverse transformation and inverse quantization, inter-prediction, intra prediction and loop
The nucleus modules such as filtering show
A data redundancy) entropy decoding: is removed using adaptive variable length coding techniques；
B) reorder: also referred to as inverse scan is mainly two-dimentional from one-dimensional transform the encoding block residual error coefficient parsed；
C) inverse transformation and inverse quantization: being replaced discrete cosine transform (DCT) with integer transform, and smallest blocks prediction is whole based on 8x8
Number dct transform matrix.Inverse transformation includes horizontal and vertical two kinds of transformation；
D) inter-prediction: correlation realizes that frame number compresses between utilizing video frame correlation instant, utilizes previous decoding image
As current encoded image with reference to figure, reference sample is selected in reference to figure, is the basic thought of inter-prediction；
E) intra prediction: for removing spatial redundancy in frame, inter-coded macroblocks code efficiency is improved.AVS technology utilizes
Frame image adjacent pixel correlation (use current block left adjacent and upper adjacent pixel is as reference pixel) realizes intra prediction；
F for eliminating blocking artifact, frame quality and code efficiency) loop filtering: are improved.Since block-based encoding and decoding are easily made
At blocking artifact.Deblocking effect is put into coding closed loop by loop filtering, to improve efficiency.
The invention mainly relates to 3 ranks or the HOSVD decomposition algorithm technologies of 4 rank tensors, it is desirable that so that decomposition coefficient tool is sparse
Structure, and HOSVD algorithm is applied in video compress, realize the compressed in layers strategy of video data, it can be in Video coding pressure
Better effects are obtained in contracting.
Tensor is number, the form expanded in institutional framework by low-dimensional to higher-dimension of vector sum matrix, it is real in the number that acquires
Strong point such as color image sequence or video flowing etc. are high dimensional data, increase data with traditional 2 rank tensors (matrix) expression
The dimension of point, while also destroying the internal structure of raw data points (frame).Tensor representation is by retaining raw data points rank (dimension
Degree) method keep data point inner structural features, such as the adjacency of pixel, contour line continuity and foreground object connectivity
With globality etc..Data storage capacity can be greatly reduced with tensor HOSVD decomposition, and is effectively denoised, that realizes video data has
Effect compression.
AVS technological deficiency and tensor resolution technical advantage:
1.AVS technology is disposable compression, and data to be compressed occupy a large amount of cachings；And tensor resolution technology can be realized in real time
Compression avoids data and occupies caching；
2, AVS technology achievees the purpose that compression using the temporal correlation of frame, and tensor resolution is under the premise of time correlation
The similarity between sequential frame image is considered simultaneously, improves compression accuracy；
3, AVS technology lacks the method for effectively searching abnormal point, and tensor resolution method carries out in fact abnormal nodes (frame)
When mark, easy-to-look-up and early warning；
4, in AVS technology, frame is saved in matrix (i.e. frame is grayscale image) form, reduces video definition and identification；
And frame is saved in tensor resolution technology with three-dimensional tensor, greatly increases image definition and identification.
The present invention is by being chronologically segmented daily video, and to per period video pictures frame automatic identification, to judge not
With the similitude of frame in period video flowing similitude and section, period video need to be retained with determination and the period retains frame number, realized
Time domain and spatial information (si) compression, reduce data-storing amount, to reduce monitoring cost, realize the optimal recovery of information；By to aobvious
The identification of work property variation node frame, timely early warning provide conveniently for monitoring personnel, avoid uninterruptedly have office hours within monitoring personnel 24 hours
Worry.
Term of the present invention is explained:
1. video compress: referring to that downscaled video data volume is under the premise of not losing useful information to reduce its occupied storage
It simultaneously improves its transmission, storage and processing speed or data is recombinated (by certain algorithm) to reduce the one of information redundancy in space
Kind technical method；
2. Real Time Compression: Information Compression prior to or be synchronized with information storage an item data compress technique, be based primarily upon
The real-time segmentation and conspicuousness node identification of video is realized；
3. tensor: a kind of organizational form and method of high dimensional data, can be used for indicate some vectors, scalar sum other
The polyteny function of linear relationship between amount, it is also possible to carry out effective expression to higher-dimension array；
4. frame tensor: every frame Picture Showing in color video is 3 rank tensor As of m × n × 3, it is by 3 matrix As
(::, 1), A (::, 2), A (::, 3) it constitutes, successively indicate the corresponding monochromatic matrix of red, green and blue, productmnFor every frame
The pixel value of picture；
5.HOSVD decompose: high order tensor singular value decomposition, this refers to by a 3(or 4) rank tensor resolution at one 3
The product of (or 4) rank core tensor and a orthogonal matrix of 3(or 4).
Citation:
[1] Li Xiao fine jade real-time AVS video decoding system [D] University of Electronic Science and Technology, 2009.
[2]H.Lu,K.N.Plataniotis, A.Venetsanopoulos.Multilinear Subspace
Learning: Dimensionality Reduction of Multidimensional Data[D].CRC Press,
2013.
Summary of the invention
The object of the present invention is to overcome the problems of the prior art, provides a kind of based on HOSVD algorithm video compress
System and method.
To realize above-mentioned technical purpose and the technique effect, the invention is realized by the following technical scheme:
One kind being based on HOSVD algorithm video compression system and method, which includes: camera device and phase
Inductive sensing device, the camera device and respective sensor are connected separately with CPLD processor and FIFO memory, the CPLD
Processor connects DSP digital signal processor with FIFO memory, and the CPLD processor is compiled with FIFO memory and AVS respectively
Code device is connected, and the AVS encoder is connected with ROM video memory；
The video-frequency compression method the following steps are included:
Step 1) building CPLD processor is set by MATLAB Integrated Development software platform by schematic diagram and MATLAB
Meter identification and compress technique algorithm, generate corresponding file destination, code are transmitted in CPLD processor chips, realize design
Digital display circuit；
Step 2 camera device and respective sensor acquisition image information are simultaneously stored in FIFO memory, and CPLD processor will
Collected pixel data is sent into the DSP digital signal processor based on MATLAB from FIFO memory, is ready for base
In the HOSVD algorithm signal processing of MATLAB.
Step 3) DSP carries out HOSVD decomposition to original video stream and data recombination is compressed, and removes time redundancy, realizes simultaneously
Prospect, background separation；
Video is streamed to AVS encoder by DSP serial communication module and carries out further coding compression by step 4), most
It is stored in ROM memory eventually.
Further, HOSVD is decomposed and is used high order tensor part singular value decomposition in the step 3), with sparse tensor table
Show, balance is reached between computation complexity and data compression.
Further, prospect and background separation realize that data are further pressed in such a way that low-rank approaches in the step 3)
Contracting and denoising.
Further, in the step 3) tensor resolution carries out in frame simultaneously by the way of to whole video Traffic Decomposition
Processing and interframe processing, often collect a frame image in camera device and respective sensor, the DSP number letter based on MATLAB
This frame image and existing frame are carried out similitude comparison and delete the high extra frame of similarity by number processor, remain with difference frame,
It is restored again into an interim flash storage, carries out next image frame grabber, continue above step, terminate until the setting period.
Further, the DSP digital signal processor is circumscribed with RAM memory and flash storage as interim storage
Device carries out data buffer storage storage, for calling the frame information closed on when HOSVD is decomposed.
The beneficial effects of the present invention are:
1, time domain and spatial redundancy Information Compression are realized with HOSVD resolution theory, maximum compression data-storing amount,
Realize the effective preservation and minimum information loss of color video；
2, high order tensor storage and HOSVD decompose the effective preservation that ensure that color video data；
3, preceding background separation and frame tensor similarity-rough set are conducive to abnormal point lookup and video stream data compression；
4, Real Time Compression facilitates timely Dynamic Discovery conspicuousness node, and automatic alarm, provides conveniently for monitoring personnel,
Avoid 24 hours worries uninterruptedly having office hours of monitoring personnel.
Detailed description of the invention
Fig. 1 (a) is the facial image saved with 3 rank tensors;
Fig. 1 (b) is the image sequence generated through Gabor filtering；
Fig. 1 (c) is the brain structure chart saved with 4 rank tensors；
Fig. 2 is system structure diagram of the invention.
Specific embodiment
It is below with reference to the accompanying drawings and in conjunction with the embodiments, next that the present invention will be described in detail.
Fig. 1 (a) is a width 3D facial image；(b) image sequence to be generated through Gabor filtering；And (c) it is video capture
Has 3D effect brain structure chart；They are successively with the preservation of 3 ranks, 3 ranks and 4 rank tensors；
Following formula shows tensor HOSVD decomposition principle:
Wherein A is N rank tensor, and the right S is core tensor (generally sparse),Expression acts on i-th of dimension of A
On an orthogonal matrix (projection).
Fig. 2 is a kind of based on HOSVD algorithm video compression system and method, which includes: that camera is set
Standby and respective sensor, the camera device and respective sensor are connected separately with CPLD processor and FIFO memory, institute
State CPLD processor and connect DSP digital signal processor with FIFO memory, the CPLD processor respectively with FIFO memory
It is connected with AVS encoder, the AVS encoder is connected with ROM video memory；
The video-frequency compression method the following steps are included:
Step 1) building CPLD processor is set by Integrated Development software platform by schematic diagram or hardware description language
Meter identification and isolation technics algorithm, generate corresponding file destination, code are transmitted in CPLD processor chips, realize design
Digital display circuit；
Step 2 camera device and respective sensor acquisition image information are simultaneously stored in FIFO memory, CPLD processor
Collected pixel data is sent into DSP digital signal processor from FIFO memory, is ready for based on tensor algorithm
Signal processing.
Step 3) DSP digital signal processor carries out tensor resolution to original video stream, data recombination is compressed, and removes on the time
Redundancy field, and by prospect and background separation；
Video is streamed to AVS encoder by the serial communication module of DSP digital signal processor and carried out by step 4)
Further coding compression, is finally stored in ROM memory.
The 4 rank tensor part singular value decompositions that HOSVD algorithm is generated using the frame sequence to timely update in the step 3)
The first principal component of frame set is extracted, background is generated, realizes prospect and background separation, prospect sequence generates sparse tensor, if setting
It is standby rotatable, then background sequence is generated, and synthesize panorama background.
The prospect sequence tensor after separation is approached using low-rank in the step 3).
Input frame (time interval can be 1s or 5s) is calculated in the step 3) and has saved consecutive frame (tensor) similarity,
Determine whether to save the input frame (deposit temporary storage), carries out next image frame grabber, realize Real Time Compression.
The DSP digital signal processor is circumscribed with RAM memory and flash storage as temporary storage, carries out
Data buffer storage storage, for calling the frame information closed in tensor resolution.
The foregoing is only a preferred embodiment of the present invention, is not intended to restrict the invention, for the skill of this field
For art personnel, the invention may be variously modified and varied.All within the spirits and principles of the present invention, made any to repair
Change, equivalent replacement, improvement etc., should all be included in the protection scope of the present invention.
Claims (4)
1. one kind is based on HOSVD algorithm video compression system, which is characterized in that the video compression system includes: camera device
And respective sensor, the camera device and respective sensor are connected separately with CPLD processor and FIFO memory, described
CPLD processor connects DSP digital signal processor with FIFO memory, the CPLD processor respectively with FIFO memory and
AVS encoder is connected, and the AVS encoder is connected with ROM video memory；
The video compression system the following steps are included:
Step 1) constructs CPLD processor, by MATLAB Integrated Development software platform, is known by schematic diagram and MATLAB design
Not with compress technique algorithm, corresponding file destination is generated, code is transmitted in CPLD processor chips, realizes the number of design
Type families system；
Step 2 camera device and respective sensor acquisition image information are simultaneously stored in FIFO memory, and CPLD processor will acquire
To pixel data from FIFO memory be sent into the DSP digital signal processor based on MATLAB in, be ready for being based on
The HOSVD algorithm signal processing of MATLAB；
Step 3) DSP carries out HOSVD decomposition and data recombination to original video stream and compresses, and removes time redundancy, at the same realize prospect,
Background separation；
Video is streamed to AVS encoder by DSP serial communication module and carries out further coding compression by step 4), is finally deposited
It is put into ROM memory；
HOSVD, which is decomposed, in the step 3) uses high order tensor part singular value decomposition, multiple calculating with sparse tensor representation
Reach balance between miscellaneous degree and data compression.
2. according to claim 1 be based on HOSVD algorithm video compression system, which is characterized in that before in the step 3)
Scape and background separation are in such a way that low-rank approaches.
3. according to claim 1 be based on HOSVD algorithm video compression system, which is characterized in that opened in the step 3)
Amount is decomposed using to whole video Traffic Decomposition by the way of, i.e., carries out processing and interframe in frame simultaneously and handle, in camera device and
Respective sensor often collects a frame image, based on the DSP digital signal processor of MATLAB by this frame image and existing frame into
Row similitude compares and deletes the high extra frame of similarity, remains with difference frame, is restored again into an interim flash storage,
Next image frame grabber is carried out, above step is continued, is terminated until the setting period.
4. according to claim 3 be based on HOSVD algorithm video compression system, which is characterized in that the DSP digital signal
Processor is circumscribed with RAM memory and flash storage as temporary storage, data buffer storage storage is carried out, in HOSVD
The frame information closed on is called when decomposition.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510677893.8A CN105357523B (en) | 2015-10-20 | 2015-10-20 | One kind being based on HOSVD algorithm video compression system and method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510677893.8A CN105357523B (en) | 2015-10-20 | 2015-10-20 | One kind being based on HOSVD algorithm video compression system and method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105357523A CN105357523A (en) | 2016-02-24 |
| CN105357523B true CN105357523B (en) | 2019-02-19 |
Family
ID=55333371
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510677893.8A Expired - Fee Related CN105357523B (en) | 2015-10-20 | 2015-10-20 | One kind being based on HOSVD algorithm video compression system and method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105357523B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20180058019A (en) * | 2016-11-23 | 2018-05-31 | 한화에어로스페이스 주식회사 | The Apparatus For Searching Image And The Method For Storing Data And The Apparatus For Storing Data |
| CN106961575A (en) * | 2017-02-24 | 2017-07-18 | 深圳汇创联合自动化控制有限公司 | A kind of efficient video monitoring system |
| CN107155111B (en) * | 2017-06-05 | 2020-02-18 | 李益永 | Video compression method and device |
| CN107507253B (en) * | 2017-08-15 | 2020-09-01 | 电子科技大学 | Multi-attribute body data compression method based on high-order tensor approximation |
| CN107528672A (en) * | 2017-09-05 | 2017-12-29 | 北京航空航天大学 | A kind of efficient wireless data transceiving method and device |
| CN107886560B (en) * | 2017-11-09 | 2021-05-25 | 网易（杭州）网络有限公司 | Animation resource processing method and device |
| CN111866443A (en) * | 2019-04-25 | 2020-10-30 | 黄河 | Video stream data storage method, device, system and storage medium |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101957909B (en) * | 2009-07-15 | 2012-09-05 | 青岛科技大学 | Digital signal processor (DSP)-based face detection method |
| KR101630303B1 (en) * | 2010-02-02 | 2016-06-14 | 삼성전자주식회사 | Apparatus for processing digital image and thereof method |
| CN102223520A (en) * | 2011-04-15 | 2011-10-19 | 北京易子微科技有限公司 | Intelligent face recognition video monitoring system and implementation method thereof |
| CN102970546B (en) * | 2012-12-13 | 2015-10-07 | 中国航空无线电电子研究所 | Video encoding unit and its implementation |
-
2015
- 2015-10-20 CN CN201510677893.8A patent/CN105357523B/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN105357523A (en) | 2016-02-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105357523B (en) | One kind being based on HOSVD algorithm video compression system and method | |
| CN108347612B (en) | Monitoring video compression and reconstruction method based on visual attention mechanism | |
| CN111355956B (en) | Deep learning-based rate distortion optimization rapid decision system and method in HEVC intra-frame coding | |
| CN104378643B (en) | A kind of 3D video depths image method for choosing frame inner forecast mode and system | |
| CN106254868B (en) | Code rate controlling method for video coding, apparatus and system | |
| CN104349074B (en) | Method, apparatus and system for generating combined digital video sequences | |
| US20160050440A1 (en) | Low-complexity depth map encoder with quad-tree partitioned compressed sensing | |
| CN107027025B (en) | A kind of light field image compression method based on macro block of pixels adaptive prediction | |
| CN106960416A (en) | A kind of video satellite compression image super-resolution method of content complexity self adaptation | |
| CN110830803B (en) | Image compression method combining block matching and string matching | |
| Gao et al. | Digital retina: A way to make the city brain more efficient by visual coding | |
| CN103826125B (en) | Concentration analysis method and device for compression monitor video | |
| CN109842799A (en) | The intra-frame prediction method and device of color component | |
| CN109982071A (en) | The bis- compression video detecting methods of HEVC based on time space complexity measurement and local prediction residual distribution | |
| CN104702959B (en) | A kind of intra-frame prediction method and system of Video coding | |
| CN115761618A (en) | Key site security monitoring image identification method | |
| CN101237581A (en) | H.264 compression domain real time video object division method based on motion feature | |
| Wu et al. | Memorize, then recall: a generative framework for low bit-rate surveillance video compression | |
| CN105930814A (en) | Method for detecting personnel abnormal gathering behavior on the basis of video monitoring platform | |
| CN103347170A (en) | Image processing method used for intelligent monitoring and high-resolution camera applied in image processing method | |
| CN103533353B (en) | A kind of near video coding system | |
| Fang et al. | Detection of HEVC double compression with different quantization parameters based on property of DCT coefficients and TUs | |
| CN110427904B (en) | Mall monitoring system, method and device based on pedestrian re-identification | |
| CN106791871B (en) | A kind of hiding detection method of motion vector modulation intelligence | |
| CN108933942A (en) | A kind of filtering method compressing video and the filter for compression video |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20190219 Termination date: 20191020 |