CN105451023B - A kind of Video Storage System and method of motion perception - Google Patents
A kind of Video Storage System and method of motion perception Download PDFInfo
- Publication number
- CN105451023B CN105451023B CN201510816617.5A CN201510816617A CN105451023B CN 105451023 B CN105451023 B CN 105451023B CN 201510816617 A CN201510816617 A CN 201510816617A CN 105451023 B CN105451023 B CN 105451023B
- Authority
- CN
- China
- Prior art keywords
- motion
- module
- video
- data
- motion perception
- 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
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N5/00—Details of television systems
- H04N5/14—Picture signal circuitry for video frequency region
- H04N5/21—Circuitry for suppressing or minimising disturbance, e.g. moiré or halo
- H04N5/213—Circuitry for suppressing or minimising impulsive noise
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Compression Or Coding Systems Of Tv Signals (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a kind of Video Storage System of motion perception and method, which includes receiving module, motion perception module, storage execution module and retrieval module.Wherein, receiving module is responsible for receiving media data to be stored from exterior；Motion perception module is responsible for analyzing the data received, including extraction motion vector, planning, low-pass filtering, part correction, medium filtering and polymerization, and exports analysis result；Execution module is stored according to above-mentioned analysis result, storage action is executed to media data；Retrieval module is responsible for the request outside response system, exports request results.The present invention differentiates whether video data is redundant data, is selectively stored, improves the utilization rate of video memory space by the resistance to vibration index of video data.Computational complexity of the present invention is low, without to video data decoding, you can calculate the resistance to vibration index of video data.
Description
Technical field
The present invention relates to the Video Storage System of motion perception and methods, belong to technical field of video image processing.
Background technology
Video Surveillance Industry rapidly develops in recent years, from analog to digital, then arrives high definition, even 4K, the clarity of image
It is continuously improved, while Video Applications explosive growth, they also bring challenges to the storage of video.
In Video Applications, often there is a large amount of redundant data (such as static picture for a long time in original video data
Face), user's video data proportion of concern comprising movable information is smaller.Video data is when being transmitted and storing
Compressed encoding need to be passed through, decoding and video content analysis computational complexity are high, and the computing capability of storage system is difficult to meet.It is existing
Some Video Storage Systems still do not account for the content of video using the video data that direct recording receives, and cause a large amount of
Waste of storage space in the storage of redundant data.And the present invention can well solve problem above.
Invention content
Present invention aims at solving, video memory space utilization rate is low, and the high problem of image analysis complexity provides
The Video Storage System and method of a kind of motion perception, the system can extract the movement of video data in the compression domain of video
Information analyzes the motion feature of video pictures content with lower operand, differentiates whether current video data belongs to redundant digit
According to, to selectively be stored, raising memory space utilization rate.
The technical scheme adopted by the invention to solve the technical problem is that：A kind of Video Storage System of motion perception, should
System includes receiving module, motion perception module, storage execution module and retrieval module.Wherein, receiving module is responsible for from system
Outside receives media data to be stored；Motion perception module is responsible for analyzing the data received, including extraction movement
Vector, planning, low-pass filtering, part correction, medium filtering and polymerization, and export analysis result；Execution module is stored according to upper
Analysis result is stated, storage action is executed to media data；Retrieval module is responsible for the request outside response system, output request knot
Fruit.
The present invention also provides a kind of implementation method of the Video Storage System of motion perception, this method includes following step
Suddenly：
Step 1：Receiving module obtains media data to be stored from the system external world, and data are transferred to motion perception respectively
Module and storage execution module.
Step 2：Motion perception module extracts the motion vector information of each macro block of media data first, and is normalized to macro
The consistent motion vector information of block size is convenient for subsequent processing.
Step 3：Motion perception module carries out noise reduction and filtering to the motion vector after normalization, reduce in video data with
The error that machine noise introduces.
Step 4：Motion perception module closes on the motion state of macro block using a macro block, and carrying out part to the macro block rectifys
Just, motion vector saltus step caused by being shaken due to light in video data is eliminated.
Step 5：Data after motion perception module corrects part carry out medium filtering, eliminate the spiced salt in vision signal
Noise, while keeping constituting the profile of the movement macro block of moving object in video pictures.
Step 6：Macro block that is adjacent and there is movement polymerize by motion perception module, forms the profile of a moving object.
Step 7：Resistance to vibration index is defined, and calculates the resistance to vibration index in a time interval, calculation formula is：
Judge whether the video data in this time section belongs to redundant data according to the index, and will differentiate that result is defeated
Go out and gives storage execution module.
Step 8：The differentiation that storage execution module is exported according to motion perception module is as a result, start or stop current media number
According to storage.
The method of the present invention is applied to the Video Storage System of motion perception.
Advantageous effect：
1, the present invention differentiates whether video data is redundant data by the resistance to vibration index of video data, selectively
It is stored, improves the utilization rate of video memory space.
2, computational complexity of the present invention is low, without to video data decoding, you can calculate the resistance to vibration of video data
Index.
Description of the drawings
Fig. 1 is the system structure diagram of the present invention.
Fig. 2 is that the motion perception module of the present invention extracts motion vector schematic diagram.
Fig. 3 is that the motion perception module of the present invention normalizes motion vector schematic diagram.
Fig. 4 is that the motion perception modular movement vector of the present invention locally corrects schematic diagram.
Fig. 5 is the motion perception modular movement vector field schematic diagram of the present invention.
Fig. 6 is that the motion perception modular movement vector of the present invention polymerize schematic diagram.
Fig. 7 is the motion perception module redundancy discriminating data schematic diagram of the present invention.
Fig. 8 is that the extraction motion vector of the present invention and motion vector normalize flow chart.
Fig. 9 is motion vector noise reduction filtering of the present invention and part correction flow chart.
Specific implementation mode
The invention is described in further detail with reference to the accompanying drawings of the specification.
As shown in Figure 1, present system includes receiving module, motion perception module stores execution module and retrieval module.
Wherein, motion perception module contains normalization, low-pass filtering, part correction, medium filtering, polymerization and differentiates submodule.It connects
The data of input are sent to motion perception module and storage executive device simultaneously by module, motion perception module exports control simultaneously
Signal processed is to storing executive device.
As shown in Fig. 2, video data is in compressed encoding, video sequence image is cut into macro block, in order to remove space-time
On correlation, be required for by links such as motion prediction, transformation and entropy codings.When carrying out motion prediction to a frame image,
Each macro block can search for oneself highest match block position of similarity in reference frame image, calculate relative to upper one
The motion vector of frame zonule.A moving object on video pictures consists of a plurality of macro blocks, therefore when object on video pictures
The movement of body is formed it is embodied in the movement of macro block.Movement sensing device extracts adjacent two frame from the media data of input
Between motion vector.
As shown in figure 3, the video data received from reception device, after entropy decoding, you can obtain each macro block
Corresponding motion vector MVi.Video data often will appear 16x16,16x8,8x8,8x4,4x4 etc. when carrying out macro block division
A variety of sizes can integrate macro block as 16x16, the motion vector after merging is movement before merging for ease of calculation
Vector weighting is added.
As shown in figure 4, video is in coding, static regional imaging also will appear random noise similar to movement effects,
Its main feature is that direction is random, but amplitude com parison is small.Can threshold value be set using 2 component absolute values of motion vector as assessment parameter
It is filtered.
As shown in figure 4, the motion vector of macro block can characterize the motion state of moving object on the whole, but local macro block
There are certain noise and error, local correction is that the motion vector of detection current macro is consistent with adjacent macroblocks motion vector
Property, if its direction of the motion vector of current macro and intensity have mutation, then it is assumed that current motion vector is unreliable, and needing can not
The motion vector refinement of the motion vector neighborhood leaned on.
As shown in figure 5, the kinematic parameter of same moving target is often a smooth process in video, filtered by intermediate value
Wave can further eliminate pulse or salt-pepper noise, while keep the edge of motion outline.After processed journey, it can be obtained
The more accurate motion vector field of moving object, left side is original image in figure, and wherein automobile is being moved, marked in image right
The motion vector of macro block, it can be seen that motion vector preferably characterizes the movement of automobile.
As shown in fig. 6, a moving object often consists of a plurality of macro blocks, therefore adjacent macro block may belong to same fortune
Animal body.Therefore there will be the adjacent macro block of movement to condense together, that is, complete the segmentation of moving object, it is shared after polymerization
The size of pixel quantity is directly proportional to video pictures exercise intensity.Being covered well there are motion vector and adjacent macro block in figure
The region where automobile is covered.
As shown in fig. 7, video is generally made of image group (i.e. GOP), the figure in one time interval of an image group packet
As frame data.Whether differentiation video data is redundant data as unit of image group in the present invention.The dynamic for defining image group is strong
Index is spent, characterizes the exercise intensity of an image group video content, which is by all motion-vector magnitudes in image group
With divided by image group frame number monotonic increasing function, the index is bigger, and the exercise intensity of video content is bigger, including user is interested
The possibility of information is bigger；The value is smaller, then the image group is that the possibility of redundant data is bigger.Black rectangle pair in figure
The image group answered is the data for needing to record a video, when recording a video beginning, 2 video groups of pre-recording, and at the end of video recording, delay video recording 2
A video group, to ensure integrality.
The specific implementation process of the method for the present invention, including：
Existing storage system hardware increases motion perception module without change on software.The input terminal of the module with deposit
Reception device docking is stored up, from storage reception device reading video data；Output end is docked with storage executive device, output storage control
Signal processed.
After motion perception module reading video data, according to the flow in Fig. 8, the motion vector sequence after normalization is obtained
Row.Normalization is carried out using formula 1.
Wherein k corresponds to the weight θ of 4x4,4x8,8x8,8x16 and 16x16 sub-macroblocks respectivelykIt is k's for current macro size
Sub-macroblock set.Motion vector after normalization according in Fig. 9 flow and formula 1 carry out noise-removed filtering, the threshold value in formula 1
T acquiescences are set as 1.5.
Locally the motion vector of each strong macro block is differentiated with neighborhood macro block according to formula 2 and 3：
Formula MV_LaX, yAmplitude for macro block and the macro block in its 8 neighborhood θ that index is (x, y) deviates from value, MV_AngX, y
For the number of angle non-zero motion vectors in value, N θ.If MV_LaX, y>50% while MV_AngX, y90 ° of ＞, then judgement should
Motion vector deviates from, and existing motion vector is replaced by formula 4.
The resistance to vibration index of each picture group is provided by formula 5：
Formula | MVX, y| it is the amplitude for the macroblock motion vector that index is (x, y), F is all macro blocks of one-frame video data
Set, GOP be a picture group, K be a picture group frame number, the corresponding MV of different scenesgopThreshold value can pass through study
Algorithm obtains.
Storage device need to cache the data of nearest 2 image groups in non-video state, when receive start video recording control
After signal processed, the data of caching are first written, then real time data is written；After receiving the control signal for stopping video recording, after continuing
Enter 2 image groups, delay stops, and such as then makes an immediate response if any new control command during being delayed stopping.
Claims (2)
1. a kind of implementation method of the Video Storage System of motion perception, which is characterized in that described method includes following steps：
Step 1：Receiving module obtains media data to be stored from the system external world, and data are transferred to motion perception module respectively
With storage execution module；
Step 2：Motion perception module extracts the motion vector information of each macro block of media data first, and is normalized to macro block ruler
Very little consistent motion vector information；
Step 3：Motion perception module carries out noise reduction and filtering to the motion vector after normalization, reduces and makes an uproar at random in video data
The error that sound introduces；
Step 4：The motion state that motion perception module closes on macro block using one carries out local correction to the macro block, and elimination regards
Motion vector saltus step caused by frequency is shaken in due to light；
Step 5：Data after motion perception module corrects part carry out medium filtering, and the spiced salt eliminated in vision signal is made an uproar
Sound, while keeping constituting the profile of the movement macro block of moving object in video pictures；
Step 6：Macro block that is adjacent and there is movement polymerize by motion perception module, forms the profile of a moving object；
Step 7：Resistance to vibration index is defined, and calculates the resistance to vibration index in a time interval, calculation formula is：
Wherein | MVX, y| it is the amplitude for the macroblock motion vector that index is (x, y), F is the collection of all macro blocks of one-frame video data
It closes, GOP is a picture group, and K is the frame number of a picture group；
Judge whether the video data in this time section belongs to redundant data according to the index, and will differentiate result export to
Store execution module；
Step 8：Storage execution module differentiates according to what motion perception module exported as a result, starting or stopping current media data
Storage.
2. a kind of implementation method of the Video Storage System of motion perception according to claim 1, which is characterized in that described
Method is applied to the Video Storage System of motion perception.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510816617.5A CN105451023B (en) | 2015-11-20 | 2015-11-20 | A kind of Video Storage System and method of motion perception |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510816617.5A CN105451023B (en) | 2015-11-20 | 2015-11-20 | A kind of Video Storage System and method of motion perception |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105451023A CN105451023A (en) | 2016-03-30 |
| CN105451023B true CN105451023B (en) | 2018-10-02 |
Family
ID=55560746
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510816617.5A Active CN105451023B (en) | 2015-11-20 | 2015-11-20 | A kind of Video Storage System and method of motion perception |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105451023B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106157388B (en) * | 2016-07-06 | 2018-08-14 | 福州瑞芯微电子股份有限公司 | The driving recording method of local coding and system |
| CN106358044B (en) * | 2016-10-28 | 2017-07-21 | 济南大学 | Motion vector detection and bearing calibration and system in a kind of three-dimensional video-frequency frame per second lifting |
| CN110677722A (en) * | 2019-09-29 | 2020-01-10 | 上海依图网络科技有限公司 | Video processing method, and apparatus, medium, and system thereof |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6512537B1 (en) * | 1998-06-03 | 2003-01-28 | Matsushita Electric Industrial Co., Ltd. | Motion detecting apparatus, motion detecting method, and storage medium storing motion detecting program for avoiding incorrect detection |
| CN1848949A (en) * | 2004-12-21 | 2006-10-18 | 三星电子株式会社 | Apparatus and method for extracting object in video surveillance system |
| CN101237581A (en) * | 2008-02-29 | 2008-08-06 | 上海大学 | H.264 compression domain real time video object division method based on motion feature |
| CN101853510A (en) * | 2010-04-20 | 2010-10-06 | 上海大学 | Movement perception model extraction method based on time-space domain |
| CN102819528A (en) * | 2011-06-10 | 2012-12-12 | 中国电信股份有限公司 | Method and device for generating video abstraction |
| CN104063883A (en) * | 2014-07-07 | 2014-09-24 | 杭州银江智慧医疗集团有限公司 | Surveillance video abstract generating method based on combination of object and key frames |
| CN104243994A (en) * | 2014-09-26 | 2014-12-24 | 厦门亿联网络技术股份有限公司 | Method for real-time motion sensing of image enhancement |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100680452B1 (en) * | 2000-02-22 | 2007-02-08 | 주식회사 팬택앤큐리텔 | Method and apparatus for updating motion vector memory |
| KR100453714B1 (en) * | 2001-12-31 | 2004-10-20 | (주)펜타마이크로 | Apparatus and Method for Motion Detection in Digital Video Recording System Using MPEG Video Compression Technique |
-
2015
- 2015-11-20 CN CN201510816617.5A patent/CN105451023B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6512537B1 (en) * | 1998-06-03 | 2003-01-28 | Matsushita Electric Industrial Co., Ltd. | Motion detecting apparatus, motion detecting method, and storage medium storing motion detecting program for avoiding incorrect detection |
| CN1848949A (en) * | 2004-12-21 | 2006-10-18 | 三星电子株式会社 | Apparatus and method for extracting object in video surveillance system |
| CN101237581A (en) * | 2008-02-29 | 2008-08-06 | 上海大学 | H.264 compression domain real time video object division method based on motion feature |
| CN101853510A (en) * | 2010-04-20 | 2010-10-06 | 上海大学 | Movement perception model extraction method based on time-space domain |
| CN102819528A (en) * | 2011-06-10 | 2012-12-12 | 中国电信股份有限公司 | Method and device for generating video abstraction |
| CN104063883A (en) * | 2014-07-07 | 2014-09-24 | 杭州银江智慧医疗集团有限公司 | Surveillance video abstract generating method based on combination of object and key frames |
| CN104243994A (en) * | 2014-09-26 | 2014-12-24 | 厦门亿联网络技术股份有限公司 | Method for real-time motion sensing of image enhancement |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105451023A (en) | 2016-03-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Li et al. | A deep learning approach for multi-frame in-loop filter of HEVC | |
| CN110189378B (en) | Video processing method and device and electronic equipment | |
| CN103618900B (en) | Video area-of-interest exacting method based on coding information | |
| JP5555221B2 (en) | Method and apparatus for adaptive color model parameter estimation of object of interest | |
| IL268214B (en) | Method for image processing and video compression | |
| CN105451023B (en) | A kind of Video Storage System and method of motion perception | |
| CN107197199A (en) | A kind of intelligent monitoring and controlling device and method for tracking target | |
| CN110263699A (en) | Method of video image processing, device, equipment and storage medium | |
| US20130155228A1 (en) | Moving object detection method and apparatus based on compressed domain | |
| dos Santos et al. | CV-C3D: action recognition on compressed videos with convolutional 3d networks | |
| CN112001308A (en) | Lightweight behavior identification method adopting video compression technology and skeleton features | |
| CN111093077A (en) | Video coding method and device, electronic equipment and storage medium | |
| CN110933415B (en) | Object region detection method, device and non-transitory computer readable medium for object region detection | |
| JPWO2021095085A1 (en) | Image processing equipment, image processing system, image processing method, and image processing program | |
| Laumer et al. | Moving object detection in the H. 264/AVC compressed domain | |
| CN116233479B (en) | Live broadcast information content auditing system and method based on data processing | |
| CN104125430B (en) | Video moving object detection method, device and video monitoring system | |
| CN104616323A (en) | Space-time significance detecting method based on slow characteristic analysis | |
| Moriyama et al. | Moving object detection in HEVC video by frame sub-sampling | |
| Savcı et al. | Fire detection in H. 264 compressed video | |
| CN112861698B (en) | Compressed domain behavior identification method based on multi-scale time sequence receptive field | |
| Ghorbel et al. | Object-based Video compression using neural networks | |
| CN113674316A (en) | Video noise reduction method, device and equipment | |
| KR102178952B1 (en) | method of providing object classification for compressed video by use of syntax-based MRPN-CNN | |
| CN104486633B (en) | Video error hides method and device |
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 |