CN103150319A - CS (Client Server) framework-based feature retrieval rear video abstraction retrieval system - Google Patents
CS (Client Server) framework-based feature retrieval rear video abstraction retrieval system Download PDFInfo
- Publication number
- CN103150319A CN103150319A CN2012104883387A CN201210488338A CN103150319A CN 103150319 A CN103150319 A CN 103150319A CN 2012104883387 A CN2012104883387 A CN 2012104883387A CN 201210488338 A CN201210488338 A CN 201210488338A CN 103150319 A CN103150319 A CN 103150319A
- Authority
- CN
- China
- Prior art keywords
- background
- moving target
- information
- target
- file
- 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
Images
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Television Signal Processing For Recording (AREA)
Abstract
The invention provides a solution scheme of a CS (Client Server) framework-based feature retrieval rear video abstraction retrieval system and application thereof to intelligent safeguard systems. The method effectively solves the problem of low efficiency of a front filtering bar of the existing video abstraction system so that the availability of the abstraction system is enhanced and the timeliness is good, and meanwhile the rear filtering condition ensures that the abstraction front information generating speed is remarkably increased.
Description
Technical field
The invention belongs to computer vision field, the particularly a kind of rearmounted video frequency abstract searching system of characteristic key solution based on the C/S framework, and the application of the method in intelligent security guard.
Technical background
Along with computer network, digitized video technology develop rapidly, obtain the application of wide model based on the video monitoring of digital network, but the thing followed is, along with the anti-scope expanding day of monitoring part, the quantity of watch-dog also increases with surprising quantity, in the face of ten hundreds of watch-dogs, video main contents how to understand fast the magnanimity control point become the maximum restriction that the project such as safe city further develops.
General solution in the face of this restriction is video summarization system, the quick Play System of video difference alienation etc. at present.but due to traditional video frequency abstract solution or be there is no filtercondition, it is namely a simple moving target summary, this interference due to the frequent motion in part in practicality is difficult to play a role in practice, having occurred afterwards can be to certain area, the video frequency abstract solution that moving target under certain logical condition extracts, but these systems will will set filtercondition often before generating summary, once summary begins to generate, these conditions can't be revised, and the well-known process that generates summary is a process more consuming time, if want to change cost that filtercondition must regenerate and be also in actual applications poor efficiency very.The present invention has extracted based on the rearmounted video frequency abstract solution of the filtercondition of C/S model, the present invention is mail to client after the video frequency abstract service end is extracted the information of moving target in advance, client loads summary info dynamically according to user's search condition.Can save a large amount of time for the summary of many conditional information retrievals like this, thereby be largely the practicality that has improved video summarization system.
Summary of the invention
The objective of the invention is treatment conditions to be processed the characteristics of poor efficiency for existing video summarization system, the rearmounted video frequency abstract solution of a kind of characteristic key based on the C/S framework has been proposed, thereby improved to a great extent the treatment effeciency of video frequency abstract, increased the practicality of video frequency abstract.
In order to realize goal of the invention, the solution of employing is as follows:
Based on the basic procedure of the rearmounted video frequency abstract searching system of the characteristic key of C/S framework as shown in Figure 1.At first, named data stream is obtained in the summarization generation control information that abstract system summary client is submitted to, service end is carried out moving Object Segmentation to data stream and is processed, extract the moving target information in video flowing, and moving target is followed the tracks of fast, and each target after following the tracks of is extracted its classification information, colouring information and size information; Afterwards, take the division of tasks method based on the Background division, moving target is carried out file organization, and mail to client after packing; After client is received generated video intermediate file, file is resolved, the user passes through client input inquiry condition, then gets summary info according to resolving Summary file, is opened up at last aobvious.
Below the present invention is introduced respectively by service end and client:
1, video frequency abstract service end:
In the video frequency abstract service end, by large functional module, service end can be divided into communication module, task module, data acquisition module and administration module.
Communication module is mainly used in controlling the work such as reception parsing, summary data packing transmission of stream information.
The mission bit stream that data acquisition module is submitted to according to client connects designated equipment and obtains data message.
Task module is the main functional modules of summary service end, and it realizes the basic function that video generates.According to the basic procedure of algorithm, this module can be divided into background model maintenance module, foreground extracting module, moving target extraction module, motion target tracking module, moving target type-collection module, moving target color extraction module, moving target size extraction module, Summary file molded tissue block.
In the present invention, used typical mixed Gaussian module as the extraction module of video background, video background upgrades and carries out according to following strategy: (1) is carried out background model according to user configured fixed interval and is upgraded; (2) according to the feedback information of foreground module, the background module is forced to upgrade.
Foreground extracting module has adopted based on the statistical method of pixel Gaussian distribution carries out foreground extraction.The method hypothesis is for any pixel in image, and it must be two kinds of roles on time shaft: prospect, background.For the security protection scene, no matter this pixel changes to background from prospect, or from the background to the prospect, the gray-scale value of pixel all has a larger variation, the present invention's hypothesis is for any pixel in video image, when its role is prospect or background (when character does not change), this pixel is in the time range of same nature, the variation service Gaussian distribution of pixel value, when the variation of pixel value is not served from Gaussian distribution, think that change has occured role's character of this pixel.The present invention utilizes slip window sampling to extract display foreground according to above hypothesis.Simultaneously, the prospect of extracting is carried out largest connected regional determination, if find that the area of a certain prospect connected region has surpassed a certain fixed threshold T, thinks that video scene has been subject to interference, this information is fed back in the background model module, forced context update to be processed.
The moving target extraction module adopts the method for traditional largest connected extracted region of foreground picture to realize.
Fast-moving target tracing overlap area-method maximum according to the consecutive frame motion followed the tracks of fast, the method supposes that namely the prospect of target two frames in front and back that coexist has partially overlapping over threshold value T, moving target sequence and the historical movement target of current extraction are mated one by one, if the right threshold value T that surpasses of sport foreground Regional Gravity And, think this target current goal of historical target for this reason, upgrade this historical target, otherwise continue, if current goal is failed and historical object matching, think that current goal is a fresh target.
Moving target type-collection module is characterized as feature pool with HUG, adopts the random forest sorter, realizes the classification of type of sports people car.
In moving target color extraction module, system's model one cover Standard Colors scale, this scale only has specific N kind color, as red, yellow, blue, green, orange etc.Every kind of color of this scale is set up the HS color statistical graph module in corresponding hsv color space, extract simultaneously moving target HS color statistic histogram in the hsv color space, every kind of color histogram of the histogram of moving target and Standard Colors scale is carried out the contrast that distributes based on card side, the colouring information of definition current goal is the immediate color of color in scale.
In moving target size extraction module, decide the moving target size with minimum edge and the maximal side of moving target.
In the present invention, the Summary file molded tissue block is comparatively crucial module, in system, adopts three grades of file structures to organize the video frequency abstract file.The first order is task level, and namely video frequency abstract is take the Video processing task of user's submission as elementary cell.The second level is background summary level, the method that adopts self-adaptation to divide at the corresponding levels is carried out the Summary file tissue, the method is followed following rule: (1) every sets up a task, and acquiescence is set up first Summary file module, the background image take the current background image as this Summary file; (2) in this task, when if the background image after finding to upgrade has larger difference than the current background image, encapsulate current Summary file module, set up new video file module based on upgrading rear backdrop, adopt the contrast that realizes the Background differentiation based on the method for hsv color space HS Color Statistical Histogram correlation pairing comparision in system.The third level is target level, the image information of the sequence of this grade file structure preservation moving target and moving target type, color, size information.Every grade of file structure is carried out the management of subordinate's file structure with the mode of XML concordance list in system.When the user shifts to an earlier date task summary when finishing dealing with, service end mails to client by communication module after with the compression of task Summary file module.The video frequency abstract file structure as shown in Figure 2.
2, video frequency abstract client:
The video frequency abstract client according to large functional module, can be divided into communication module, administration module, summary parsing and information filtering module, summary playing module, the corresponding source document playback of summary and transmit control module.
Communication module realizes controlling stream information transmission, the reception of service end status information, the reception of video frequency abstract file module and decompress(ion).
Summary is resolved and the information filtering module can specifically be divided into video frequency abstract filtering information interpolation module, video frequency abstract document analysis module, summarized radio molded tissue block.
The module that arranges the video frequency abstract filtering information realizes that mainly the client arranges current filtering information by this module, namely extracts the summary in specified type, color, size, zone and plays.
Video frequency abstract document analysis module is resolved the video frequency abstract file module that service end sends, at first the task ID information of submitting to according to the user, navigate to the task Summary file, video frequency abstract management index file in simulation task Summary file is to traveling through each based on the video frequency abstract file of background, for any one video frequency abstract file, according to the moving target index file in this document, traversal target information, combined filtering information extraction go out the target travel target and are committed to the summarized radio molded tissue block.
The summarized radio molded tissue block is according to the video frequency abstract that receives, and the summarized radio that carries out take Background as unit generates, and is committed to the video frequency abstract playing module.
The video frequency abstract playing module carries out broadcast and the corresponding Play Control of Summary file according to the video frequency abstract that receives.And the response user is to the query event of target, the target location of clicking according to the user corresponds to target information, finally obtain original video files information corresponding to this target, original video files information is submitted to the corresponding source document playback of summary and transmit control module, and realizing makes a summary back looks into playing function.
Make a summary corresponding source document playback and transmit control module realizes that original video is play and corresponding Play Control function.
Above method proposed by the invention, more traditional video abstraction generating method can generate summary more fast, and can realize the quick-replaceable of search condition, thus the practicality of the video summarization system that improves to a great extent.
Description of drawings
Fig. 1 is basic flow sheet of the present invention;
Fig. 2 is video frequency abstract file structure figure of the present invention.
Embodiment
1, each module practical function of service end intermodule and basic procedure are described below:
Communication module realizes that control stream, the state flow information based on TCP sends, and the video frequency abstract file is based on the compression of binary word throttling, based on the Summary file sending function of File Transfer Protocol.
The main implementation algorithm task class of task module, unique information task object of each task creation, this algorithm task class realizes that mainly the video background module is safeguarded, the video foreground information extraction, moving target extracts, motion target tracking, the moving target type information extracts, the moving target colouring information extracts, and the moving target size information is extracted, the Summary file Organization and management.
Data acquisition module, the data source information that main realization provides according to task loads video stream data, and finally converts RGB frame image data stream to.
The management mould, the organizing and coordinating of function of mainly realizing each module realized the task distribution, task resource reclaims.
2, between client modules, each module practical function and basic procedure are described below:
Communication module mainly realizes receiving and sending based on the control stream information of Transmission Control Protocol and state flow information, based on the video frequency abstract file reception of File Transfer Protocol, based on the Summary file solution compression functions of binary file stream.
Administration module is mainly realized coordination and the management between each functions of modules.
Summary is resolved and the information filtering module realizes the parsing of video frequency abstract file module, and the filtering information setting is filtered moving target in summary, function of the abstract after generation is filtered.
The summary playing module is mainly realized the broadcast of video frequency abstract, and corresponding Play Control function, video frequency abstract target playback acquisition of information function.
Make a summary corresponding source document playback and transmit control module mainly realize making a summary corresponding video file remote playing and corresponding Play Control function.
Claims (6)
1. one kind based on the rearmounted video frequency abstract searching system of the characteristic key of C/S framework, it is characterized in that system scheme takes the C/S framework to carry out system organization, system realizes video data acquiring at server end, moving target extracts and moving target information extraction, data compression; Realizing in client that the goal filtering, moving target of packet decompress(ion), based on motion target information are play represents and corresponding playing back videos represents.
2. method described according to claim 1, it is characterized in that more new solution of system server terminal moving target extraction employing background modeling and mobilism, it shows, at first background dynamics upgrades the context update based on fixed frequency, next is to feed back context update and foreground detection related, if the prospect maximum area during greater than the certain proportion of whole scene, is forced context update, background model uses mixed Gauss model to carry out background modeling; Foreground extraction adopts the statistics side based on the pixel Gaussian distribution to extract; The motion target tracking method adopts the area coincidence method to follow the tracks of fast.
3. method described according to claim 1, it is characterized in that system server carries out the moving target that extracts based on kind, color and big or small moving target information extraction, at first it show as and will carry out target being divided into " vehicle " and " people " based on the target classification algorithm that improves the HOG feature to moving target; Secondly, system utilizes the algorithm of color-based model to carry out the extraction of color target information; At last, the minimum length of side and the maximal side of record object in whole scene retrieved foundation as the size of target in carrying out the analysis of target traversal.
4. method described according to claim 1, is characterized in that utilizing the division of tasks method of dividing based on Background, and it shows as long task is carried out based on the rational division of background, and with this foundation as local Summary file tissue; So-called background is rationalized division and is referred to, if one section original video has comprised some sections larger video segments of background difference, this section video is carried out cluster take the difference of background image as classification foundation, and carry out the local file tissue of task video take each class as unit, these local files have such feature: comprise unique background image in file, comprise some moving target foreground images that background therewith is associated in time in file; The background summary of this file structure called after task summary is cut apart, and cut apart foundation with the index file of task name for the background summary of task summary.
5. method according to claim 4, the background summary that it is characterized in that the task summary is cut apart index file and had following content: at first, this index file comprises the appointed task relevant information, comprises temporal information, fileinfo; Secondly, index file records the essential information of moving target take moving target as the unit, comprises temporal information, classification information, size information, positional information.
6. method according to claim 1, it is characterized in that system client is made a summary based on following flow process to represent and playback: at first, the background summary of the task summary that client loading service end is set up is cut apart index; Secondly, client is cut apart directory entry according to the search condition of user's input to the background summary and is filtered, and the entry after filtration is final result for retrieval; At last, the directory entry after filtering according to retrieval navigates to moving target, and plays and represent.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210488338.7A CN103150319B (en) | 2012-11-16 | 2012-11-16 | The rearmounted video frequency abstract searching system of a kind of characteristic key based on CS framework |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210488338.7A CN103150319B (en) | 2012-11-16 | 2012-11-16 | The rearmounted video frequency abstract searching system of a kind of characteristic key based on CS framework |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103150319A true CN103150319A (en) | 2013-06-12 |
| CN103150319B CN103150319B (en) | 2016-06-29 |
Family
ID=48548400
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210488338.7A Active CN103150319B (en) | 2012-11-16 | 2012-11-16 | The rearmounted video frequency abstract searching system of a kind of characteristic key based on CS framework |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103150319B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104427369A (en) * | 2013-09-09 | 2015-03-18 | 联想(北京)有限公司 | Remote terminal equipment, remotely controlled terminal equipment and method used thereby |
| CN104581437A (en) * | 2014-12-26 | 2015-04-29 | 中通服公众信息产业股份有限公司 | Video abstract generation and video backtracking method and system |
| CN104754248A (en) * | 2013-12-30 | 2015-07-01 | 浙江大华技术股份有限公司 | Method and device for acquiring target snapshot |
| CN109165317A (en) * | 2018-10-31 | 2019-01-08 | 杭州恒生数字设备科技有限公司 | A kind of real time monitoring aspect indexing inquiry system |
| CN110674347A (en) * | 2019-09-02 | 2020-01-10 | 南京邮电大学 | Visual shielding double-layer AP video abstract generation method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050105002A1 (en) * | 2002-01-31 | 2005-05-19 | Matsushita Electric Industrial Co., Ltd | Digest video specification system, digest video providing system, digest video specifying method, digest video providing method, and medium and program therefor |
| US20090119704A1 (en) * | 2004-04-23 | 2009-05-07 | Koninklijke Philips Electronics, N.V. | Method and apparatus to catch up with a running broadcast or stored content |
| US20100125581A1 (en) * | 2005-11-15 | 2010-05-20 | Shmuel Peleg | Methods and systems for producing a video synopsis using clustering |
| CN101753853A (en) * | 2009-05-13 | 2010-06-23 | 中国科学院自动化研究所 | Fusion method for video scene segmentation |
| CN102156707A (en) * | 2011-02-01 | 2011-08-17 | 刘中华 | Video abstract forming and searching method and system |
-
2012
- 2012-11-16 CN CN201210488338.7A patent/CN103150319B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20050105002A1 (en) * | 2002-01-31 | 2005-05-19 | Matsushita Electric Industrial Co., Ltd | Digest video specification system, digest video providing system, digest video specifying method, digest video providing method, and medium and program therefor |
| US20090119704A1 (en) * | 2004-04-23 | 2009-05-07 | Koninklijke Philips Electronics, N.V. | Method and apparatus to catch up with a running broadcast or stored content |
| US20100125581A1 (en) * | 2005-11-15 | 2010-05-20 | Shmuel Peleg | Methods and systems for producing a video synopsis using clustering |
| CN101753853A (en) * | 2009-05-13 | 2010-06-23 | 中国科学院自动化研究所 | Fusion method for video scene segmentation |
| CN102156707A (en) * | 2011-02-01 | 2011-08-17 | 刘中华 | Video abstract forming and searching method and system |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104427369A (en) * | 2013-09-09 | 2015-03-18 | 联想(北京)有限公司 | Remote terminal equipment, remotely controlled terminal equipment and method used thereby |
| CN104427369B (en) * | 2013-09-09 | 2018-08-10 | 联想(北京)有限公司 | Remote control end equipment is remotely-controlled end equipment and the method for it |
| CN104754248A (en) * | 2013-12-30 | 2015-07-01 | 浙江大华技术股份有限公司 | Method and device for acquiring target snapshot |
| CN104754248B (en) * | 2013-12-30 | 2018-05-01 | 浙江大华技术股份有限公司 | A kind of method and device for obtaining target snapshot |
| CN104581437A (en) * | 2014-12-26 | 2015-04-29 | 中通服公众信息产业股份有限公司 | Video abstract generation and video backtracking method and system |
| CN104581437B (en) * | 2014-12-26 | 2018-11-06 | 中通服公众信息产业股份有限公司 | A kind of video frequency abstract generates and the method and system of video backtracking |
| CN109165317A (en) * | 2018-10-31 | 2019-01-08 | 杭州恒生数字设备科技有限公司 | A kind of real time monitoring aspect indexing inquiry system |
| CN110674347A (en) * | 2019-09-02 | 2020-01-10 | 南京邮电大学 | Visual shielding double-layer AP video abstract generation method |
| CN110674347B (en) * | 2019-09-02 | 2022-04-01 | 南京邮电大学 | Visual shielding double-layer AP video abstract generation method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103150319B (en) | 2016-06-29 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110147722A (en) | A kind of method for processing video frequency, video process apparatus and terminal device | |
| CN103150319B (en) | The rearmounted video frequency abstract searching system of a kind of characteristic key based on CS framework | |
| CN202003350U (en) | Video summary system | |
| CN102724385B (en) | A kind of Intelligent video analysis method and device | |
| CN111310041B (en) | Image-text publishing method, model training method and device and storage medium | |
| CN105045820A (en) | Method for processing video image information of mass data and database system | |
| CN102903124A (en) | Moving object detection method | |
| CN103118220B (en) | A kind of Key-frame Extraction Algorithm based on multidimensional characteristic vectors | |
| CN102968637A (en) | Complicated background image and character division method | |
| CN103778237A (en) | Video abstraction generation method based on space-time recombination of active events | |
| CN103177436B (en) | Method and system for tracking object | |
| CN103929685A (en) | Video abstract generating and indexing method | |
| CN103927364A (en) | Storage method and system and display system for video abstract data | |
| CN112332981B (en) | Data processing method and device | |
| CN110166851B (en) | Video abstract generation method and device and storage medium | |
| CN106488257A (en) | A kind of generation method of video file index information and equipment | |
| CN112817755A (en) | Edge cloud cooperative deep learning target detection method based on target tracking acceleration | |
| CN112364176A (en) | Method, equipment and system for constructing personnel action track | |
| CN112653749A (en) | Edge computing-based complex event processing system and method for Internet of things | |
| CN107301245A (en) | A kind of power information video searching system | |
| CN104333732A (en) | Distributed video analysis method and system thereof | |
| CN114222163A (en) | Intelligent video analysis system for business hall | |
| CN112306985A (en) | Digital retina multi-modal feature combined accurate retrieval method | |
| CN114513681A (en) | Video processing system, method, device, electronic equipment and storage medium | |
| CN114449346A (en) | Video processing method, device, equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| DD01 | Delivery of document by public notice | ||
| DD01 | Delivery of document by public notice |
Addressee: PCI-Suntektech Co., Ltd. Document name: Notification of Passing Examination on Formalities |
|
| CP03 | Change of name, title or address | ||
| CP03 | Change of name, title or address |
Address after: Room 306, area 2, building 1, Fanshan Venture Center, Panyu energy saving science and Technology Park, 832 Yingbin Road, Donghuan street, Panyu District, Guangzhou, Guangdong 510000 Patentee after: Jiadu Technology Group Co.,Ltd. Address before: No.4, Jiangong Road, Tianhe Software Park, Guangzhou, Guangdong 510665 Patentee before: PCI-SUNTEKTECH Co.,Ltd. |