CN102222111B - Method for retrieving high-definition video content - Google Patents
Method for retrieving high-definition video content Download PDFInfo
- Publication number
- CN102222111B CN102222111B CN2011101801770A CN201110180177A CN102222111B CN 102222111 B CN102222111 B CN 102222111B CN 2011101801770 A CN2011101801770 A CN 2011101801770A CN 201110180177 A CN201110180177 A CN 201110180177A CN 102222111 B CN102222111 B CN 102222111B
- Authority
- CN
- China
- Prior art keywords
- video
- information
- frame
- target
- index 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.)
- Expired - Fee Related
Links
Images
Abstract
The invention discloses a method for retrieving a high-definition video content, which is characterized by comprising the following steps of: (1) dividing the high-definition video content into a group information frame and an individual information frame initially, and storing the corresponding video frame information to establish a video index file; (2) constructing a high-level semantic library, and establishing a semantic library comprising various events and a plurality of pieces of target information by using the conventional knowledge; (3) performing video retrieval on an individual information frame according to the high-level semantic library, and establishing an individual information content index file; (4) performing the video retrieval on a group information frame according to the high-level semantic library, and establishing a group information content index file; and (5) merging the individual information content index file and the group information content index file to form an index file for the entire video content.
Description
Technical field
The present invention relates to video search technique area, specifically, relate to a kind of HD video content search method and system.
Background technology
Along with Development of Multimedia Technology and the progress of making process technology, people's obtainable video of institute quantitatively or qualitatively all is greatly improved and improves no matter be.Meanwhile, people are also improving the requirement of aspects such as the sharpness of video pictures, picture detail information, and common SD video can not satisfy people's demand.The image quality that HD video had is good, image sharpness is high, and people are needed just to highlight characteristics such as detailed information, and therefore gradually HD video will become main flow.So in the face of a large amount of HD video data the time, how people go to obtain own interested content and don't need expend the too many time to become a vital problem.Therefore, the Content-based Video Retrieval technology becomes the focus of concern,
The Content-based Video Retrieval technology is to use computer image processing technology, and the video pictures content is analyzed and distinguished, identifying the interested content of people, and the result is shown.Present Content-based Video Retrieval technology all is the retrievals to the SD video.Because SD video itself can't be taken into account the restriction of the shortcoming of whole and detailed information; Existing Content-based Video Retrieval technology can't be come out the advantage applies of HD video, promptly can't effectively extract the detailed information point that is comprised in the HD video picture.This is the weak point of prior art.
Summary of the invention
The technical matters that the present invention will solve provides a kind of HD video content search method, can carry out the efficient content retrieval to HD video, can embody the abundant characteristics of HD video detail content.
The present invention adopts following technological means to realize goal of the invention:
A kind of HD video content search method is characterized in that, may further comprise the steps:
(1) HD video content Preliminary division is community information frame and individual information frame, and the video index file is set up in the frame of video information preservation of correspondence; Individual information frame after dividing and community information frame are also preserved with the video index file that the frame of video information that is associated is set up correspondence, and the information of the described frame of video of this step comprises frame of video frame number and video frame time;
(2) structure high-level semantic storehouse utilizes existing knowledge, sets up the semantic base that includes multiple incident, a plurality of target informations; This semantic base includes the detailed knowledge of incident, target, sight etc. to be described, and the content of these descriptions will become the criterion to incident in the video content takes place, target exists and judging appears in sight;
(3) according to the high-level semantic storehouse, the individual information frame is carried out video frequency searching, set up individual information content indexing file; Judge that the target that the incident that whether has semantic base to comprise in the video content takes place, whether has semantic base to comprise exists, the sight that whether has semantic base to comprise occurs.The index that this step is set up comprises the information of interest that exists in the video content (like the shape of target, color, texture etc.), frame of video frame number and video frame time etc., and all can be used for the key element that links together with information of interest and video.
(4) according to the high-level semantic storehouse, the community information frame is carried out video frequency searching, set up community information content indexing file; The target whether incident whether judge has semantic base to comprise in the video content of colony's frame of video takes place, has semantic base to comprise exists; The sight that whether has semantic base to comprise occurs; And the target of each motion carried out ontoanalysis, set up index to analyzing the result who accomplishes.The index that this step is set up comprise the community information (like rally information, the information etc. of fighting) that exists in the video content and moving target individual information (like the shape of target, color, texture etc.), frame of video frame number and video frame time etc. all can be used for the key element that links together with information of interest and video.
(5) individual information content indexing file and community information content indexing file are merged, form index file to the whole video content.After the retrieval of accomplishing community information frame and individual information frame, with formed two independently index file merge, formation promptly includes the index file of whole colonies and individual information to the net result of HD video content analysis.
As the further qualification to the present technique scheme, said step (1) comprises the steps:
(1.1) set a moving target threshold values;
(1.2) the moving target quantity of each frame of video in the video content is analyzed, the frame of video that is lower than the target threshold values is the individual information frame of video, and the frame of video that is higher than the target threshold values is the community information frame of video;
(1.3) the individual information frame of video after will dividing, community information frame of video are set up corresponding index file and preservation with the frame of video information that is associated respectively.
As the further qualification to the present technique scheme, the frame of video information in the said step (1.3) comprises frame of video frame number and frame of video frame time.
As the further qualification to the present technique scheme, the rule in the said high-level semantic storehouse comprises the high-level semantic description of individual behavior incident and group behavior incident;
Said step (3) comprising:
(3.1) use edge detection operator to obtain the shape information of moving target to the target in the video pictures;
(3.2) use color histogram to extract the colouring information of target, convert coloured image into gray level image;
(3.3) the texture information relevant information of the feature extraction target of use grey level histogram;
(3.4) shape that the front is extracted, color and texture relevant information use the high-level semantic storehouse of structure to judge respectively, then resulting judged result are carried out and computing, if the result is the information of interest appearance for decidable very then.
Said step (4) comprising:
(4.1) at first extract community information; When a plurality of targets of appearance are assembled in a certain zone; Can these a plurality of targets be regarded as a bigger simple target; Can use the movement locus extraction algorithm to obtain the movement locus of colony to it, utilize the size characteristic of target and aggregation zone size to come the density of estimating target and the variation of monitored density, then use image partition method to be divided into single target it;
(4.2) use edge detection operator to obtain the shape information of moving target to single target; Use color histogram to extract the colouring information of target; Convert coloured image into gray level image; Use the texture information relevant information of the feature extraction target of grey level histogram; Use the high-level semantic storehouse of structure to judge respectively shape, color and texture relevant information that the front extracts, then resulting judged result is carried out and computing, if the result is the information of interest appearance for decidable very then;
(4.3) and with the information of interest that above processing obtains use the high-level semantic storehouse of the relevant information that makes up to judge that the result that judgement is obtained carries out and computing, if true, then colony's incident takes place.
Said step (5) comprising:
(5.1) index file that the individual information frame is formed is inserted into the sequencing of time in the index of community information frame and forms preliminary whole index file;
(5.2) whether include identical information with the index data that detects same video frame number then, the information if any then deletion redundancy forms final video content index file.
Compared with prior art; Advantage of the present invention and good effect are: in video content; Can be not identical to our institute's information of interest of frame of video that contains community information and the frame of video that contains individual information; Therefore resulting HD video is divided into community information frame and individual information frame, the utilization diverse ways is retrieved in follow-up video content retrieval, especially when the community information frame is retrieved; Give full play to the characteristics that HD video highlights detailed information, carry out the information excavating of a plurality of individualities.So both can accelerate the efficient of video frequency searching, again can be more obtain interested detail content.In result for retrieval, use index file that institute's information of interest and corresponding video are linked together, can consult playback easily with video.
Description of drawings
Fig. 1 is the process flow diagram of the preferred embodiment of the present invention.
Fig. 2 is the system block diagram of the preferred embodiment of the present invention.
Embodiment
Below in conjunction with accompanying drawing and preferred embodiment the present invention is made detailed description further.
Referring to Fig. 1, Fig. 2; The object that the HD video content search method of present embodiment is directed against is the marine vessel navigation channel HD video content retrieval that cruises; Certainly; Method of the present invention is not limited to this embodiment, please refer to Fig. 1, and the cruise enforcement of HD video content retrieval of marine vessel navigation channel realizes according to following flow process.
(1) ship that has the obtained HD video that cruises is carried out initial analysis, divide individual information frame and community information frame; Analyze according to the cruise complexity of time institute's information of interest, marine scene of ship sea; Can the moving target threshold values be made as 1; Be occur in the video pictures one or when being lower than a moving object this frame of video be considered to the individual information frame; If during greater than a moving target, then be regarded as the community information frame.
(2) set up the high-level semantic storehouse for the ship sea institute's information of interest of cruising; The sea institute's information of interest of cruising is more, and interested partial information is set up the high-level semantic storehouse in this example, cruising with the navigation channel.The main purpose of cruising in the navigation channel is to check the barrier that exists on navigation mark and the navigation channel in navigation channel, therefore sets up the high-level semantic storehouse with navigation mark and barrier as information of interest.To can describing from the characteristic of aspects such as its color, shape, texture of these two targets, also having the information of a concern for navigation mark is exactly the numbering of navigation mark, sets up relevant high-level semantic storehouse through the description of above several respects.Semantic base is comprising and is being used for judging whether video content has information of interest to exist; Be that the mode that end users are familiar with is set up the criterion of some judgements; In other words to the definition of some things, behavior, scene etc., and machine uses these criterions or definition can go to discern these things, behavior and scene etc. accurately.
(3) individual information frame and community information frame are carried out content analysis and set up index file; Individual information frame and community information frame are carried out content analysis; Make criterion or definition in the high-rise adopted storehouse of term judge whether have navigation mark or barrier in the video pictures; If there is navigation mark; Then the numbering with navigation mark identifies, and sets up corresponding index file for it, and relevant information and frame of video information that navigation mark and barrier are existed are associated.Because the quantity of information that is comprised in individual information frame and the community information frame is different, therefore, in order more effectively to handle video, obtains abundant content information more, uses different method for processing video frequency respectively two kinds of information frames to be handled.
(4) form in the fusion steps (three) two independently individual information frame index file and community information frame index files.Because navigation mark and barrier possibly all occur in individual information frame and community information frame; Therefore; Two whole HD videos of expression that independently index file can not be complete; Need identical part in two index files be handled, form a file that can carry out index complete HD video.
(5) content retrieval; For example navigation mark in the navigation channel is retrieved, the numbering of then searching corresponding frame of video and navigation mark through index file is shown to the user.
(6) video playback; Can form a video of simplifying for the frame of video of navigation mark and barrier existence and on video playback device, carry out playback.
In sum, the present invention can make things convenient for, and efficiently HD video is carried out content retrieval, can effectively extract the detailed information point of HD video.
The invention also discloses a kind of HD video content retrieval system, comprising:
High-definition camera control module: be responsible for the operation control of high-definition camera; High-definition camera is carried out real-time parameter setting and control; To obtain the best effect of video camera; Can carry out alternately with system control module; The result that instruction that the executive system control module issues and feedback are carried out, this module comprises that to the parameter setting of video camera aperture parameters setting, the setting of visual field parameter, color mode setting, code stream formatting, the setting of coding size, frame per second setting, bit rate setting, image quality setting, the OSD demonstration of high-definition camera are provided with, the network parameter setting.
System control module: be responsible for the overall operation of total system, can send instruction and connect them feedback after instruction is carried out to high-definition camera control module, video content analysis module, video content retrieval module.
Memory module: be responsible for preserving video data and content indexing and the result for retrieval that high-definition camera obtains, and when retrieval, content indexing and video data offered the video content retrieval module.
Video content analysis module: the video content to high-definition camera obtained is analyzed, and forms the content indexing file, and the parameter in the analytic process is set, such as the setting of moving target threshold values, the setting of semantic base.
Video content retrieval module: according to the search condition of setting the index file that video content analysis module forms is retrieved, and the result that will retrieve and the video content that is associated send to display module and show;
Display module: be responsible for the result for retrieval of retrieval module is showed the client through display terminal.
Certainly, above-mentioned explanation is not a limitation of the present invention, and the present invention also is not limited only to above-mentioned giving an example, and variation, remodeling, interpolation or replacement that those skilled in the art are made in essential scope of the present invention also belong to protection scope of the present invention.
Claims (4)
1. a HD video content search method is characterized in that, may further comprise the steps:
(1) HD video content Preliminary division is community information frame and individual information frame, and the video index file is set up in the frame of video information preservation of correspondence;
(2) structure high-level semantic storehouse utilizes existing knowledge, sets up the semantic base that includes multiple incident, a plurality of target informations;
(3) according to the high-level semantic storehouse, the individual information frame is carried out video frequency searching, set up individual information content indexing file, specific as follows:
(3.1) use edge detection operator to obtain the shape information of moving target to the target in the video pictures;
(3.2) use color histogram to extract the colouring information of target, convert coloured image into gray level image;
(3.3) the texture information relevant information of the feature extraction target of use grey level histogram;
(3.4) shape that the front is extracted, color and texture relevant information use the high-level semantic storehouse of structure to judge respectively, then resulting judged result are carried out and computing, if the result is the information of interest appearance for decidable very then;
(4) according to the high-level semantic storehouse, the community information frame is carried out video frequency searching, set up community information content indexing file, specific as follows:
(4.1) at first extract community information; When a plurality of targets of appearance are assembled in a certain zone; Can these a plurality of targets be regarded as a bigger simple target; Can use the movement locus extraction algorithm to obtain the movement locus of colony to it, utilize the size characteristic of target and aggregation zone size to come the density of estimating target and the variation of monitored density, then use image partition method to be divided into single target it;
(4.2) use edge detection operator to obtain the shape information of moving target to single target; Use color histogram to extract the colouring information of target; Convert coloured image into gray level image; Use the texture information relevant information of the feature extraction target of grey level histogram; Use the high-level semantic storehouse of structure to judge respectively shape, color and texture relevant information that the front extracts, then resulting judged result is carried out and computing, if the result is the information of interest appearance for decidable very then;
(4.3) information of interest that above processing is obtained uses the high-level semantic storehouse of the dependent event that makes up to judge that the result that judgement is obtained carries out and computing, if true, then colony's incident takes place;
(5) individual information content indexing file and community information content indexing file are merged, form index file to the whole video content, specific as follows:
(5.1) index file that the individual information frame is formed is inserted into the sequencing of time in the index of community information frame and forms preliminary whole index file;
(5.2) whether include identical information with the index data that detects same video frame number then, the information if any then deletion redundancy forms final video content index file.
2. according to the said HD video content search method of claim 1, it is characterized in that said step (1) comprises the steps:
(1.1) set a moving target threshold values;
(1.2) the moving target quantity of each frame of video in the video content is analyzed, the frame of video that is lower than the target threshold values is the individual information frame of video, and the frame of video that is higher than the target threshold values is the community information frame of video;
(1.3) the individual information frame of video after will dividing, community information frame of video are set up corresponding index file and preservation with the frame of video information that is associated respectively.
3. according to the said HD video content search method of claim 2, it is characterized in that the frame of video information in the said step (1.3) comprises frame of video frame number and frame of video frame time.
4. according to the said HD video content search method of claim 1, it is characterized in that the rule in the said high-level semantic storehouse comprises the high-level semantic description of individual behavior incident and group behavior incident.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011101801770A CN102222111B (en) | 2011-06-30 | 2011-06-30 | Method for retrieving high-definition video content |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2011101801770A CN102222111B (en) | 2011-06-30 | 2011-06-30 | Method for retrieving high-definition video content |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102222111A CN102222111A (en) | 2011-10-19 |
| CN102222111B true CN102222111B (en) | 2012-10-31 |
Family
ID=44778663
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2011101801770A Expired - Fee Related CN102222111B (en) | 2011-06-30 | 2011-06-30 | Method for retrieving high-definition video content |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102222111B (en) |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103020260A (en) * | 2012-12-24 | 2013-04-03 | 中国科学院半导体研究所 | Video query method |
| CN103198110A (en) * | 2013-03-28 | 2013-07-10 | 广州中国科学院软件应用技术研究所 | Method and system for rapid video data characteristic retrieval |
| CN103279493A (en) * | 2013-05-03 | 2013-09-04 | 吴军 | Intelligent mass surveillance video analysis system |
| CN103279580B (en) * | 2013-06-24 | 2016-05-11 | 魏骁勇 | Based on the video retrieval method of novel semantic space |
| CN107315984B (en) * | 2016-04-26 | 2020-04-28 | 杭州华为企业通信技术有限公司 | Pedestrian retrieval method and device |
| CN106326382A (en) * | 2016-08-16 | 2017-01-11 | 广东中星电子有限公司 | Video data storage and search method and device based on layered code stream |
| CN107517405A (en) * | 2017-07-31 | 2017-12-26 | 努比亚技术有限公司 | The method, apparatus and computer-readable recording medium of a kind of Video processing |
| CN108337482A (en) * | 2018-02-08 | 2018-07-27 | 北京信息科技大学 | The storage method and system of monitor video |
| CN110209879B (en) * | 2018-08-15 | 2023-07-25 | 腾讯科技（深圳）有限公司 | Video playing method, device, equipment and storage medium |
| CN113361313A (en) * | 2021-02-20 | 2021-09-07 | 温州大学 | Video retrieval method based on multi-label relation of correlation analysis |
| CN114786042A (en) * | 2022-04-12 | 2022-07-22 | 北京字节跳动网络技术有限公司 | Video playing method, device, equipment and storage medium |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101739568A (en) * | 2009-11-04 | 2010-06-16 | 北京交通大学 | Layered observation vector decomposed hidden Markov model-based method for identifying behaviors |
| CN101778260A (en) * | 2009-12-29 | 2010-07-14 | 公安部第三研究所 | Method and system for monitoring and managing videos on basis of structured description |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20030079228A1 (en) * | 2001-10-23 | 2003-04-24 | Shu Lin | Searching for frames to perform a trick mode |
-
2011
- 2011-06-30 CN CN2011101801770A patent/CN102222111B/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101739568A (en) * | 2009-11-04 | 2010-06-16 | 北京交通大学 | Layered observation vector decomposed hidden Markov model-based method for identifying behaviors |
| CN101778260A (en) * | 2009-12-29 | 2010-07-14 | 公安部第三研究所 | Method and system for monitoring and managing videos on basis of structured description |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102222111A (en) | 2011-10-19 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102222111B (en) | Method for retrieving high-definition video content | |
| WO2022116888A1 (en) | Method and device for video data processing, equipment, and medium | |
| Smeaton et al. | Video shot boundary detection: Seven years of TRECVid activity | |
| CN111327945B (en) | Method and apparatus for segmenting video | |
| CN113613065B (en) | Video editing method and device, electronic equipment and storage medium | |
| CN103347167A (en) | Surveillance video content description method based on fragments | |
| JP2005032246A (en) | Image group representation method, descriptor derived by the same, search method and device, computer program, and storage medium | |
| CN111582122B (en) | System and method for intelligently analyzing behaviors of multi-dimensional pedestrians in surveillance video | |
| US11037604B2 (en) | Method for video investigation | |
| CN110619284B (en) | Video scene division method, device, equipment and medium | |
| KR20090093904A (en) | Apparatus and method for scene variation robust multimedia image analysis, and system for multimedia editing based on objects | |
| JP2015149030A (en) | Video content violence degree evaluation device, video content violence degree evaluation method, and video content violence degree evaluation program | |
| KR20190137924A (en) | Modify digital video content | |
| CN106488257A (en) | A kind of generation method of video file index information and equipment | |
| JP5116017B2 (en) | Video search method and system | |
| CN107301245A (en) | A kind of power information video searching system | |
| US20070061727A1 (en) | Adaptive key frame extraction from video data | |
| Premaratne et al. | Structural approach for event resolution in cricket videos | |
| Wang et al. | Visual saliency based aerial video summarization by online scene classification | |
| Liu et al. | Within and between shot information utilisation in video key frame extraction | |
| KR20230016394A (en) | Method and System for Generating Video Synopsis Based on Abnormal Object Detection | |
| Gao et al. | Shot-based similarity measure for content-based video summarization | |
| Gao et al. | Accelerating shot boundary detection by reducing spatial and temporal redundant information | |
| Masneri et al. | Towards semi-automatic annotation of video and audio corpora | |
| KR101477051B1 (en) | Method for object recognition through image-code transformation |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20121031 Termination date: 20160630 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |