CN105630833A - Video information slice query method - Google Patents
Video information slice query method Download PDFInfo
- Publication number
- CN105630833A CN105630833A CN201410622906.7A CN201410622906A CN105630833A CN 105630833 A CN105630833 A CN 105630833A CN 201410622906 A CN201410622906 A CN 201410622906A CN 105630833 A CN105630833 A CN 105630833A
- Authority
- CN
- China
- Prior art keywords
- camera lens
- section
- shot
- slice
- similar
- 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.)
- Pending
Links
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a video information slice query method, and belongs to the technical field of video query methods. The fast video information slice query method comprises the following steps: (1) carrying out shot segmentation by using a character string matching algorithm; (2) fusing each shot into a shot slice iq by using a time slice fusion algorithm and extracting the low-level features of the shot slice; (3) circulation: for each shot slice, searching k similar shot slices similar to the shot slice in a shot slice library, and filtering out the unrelated shots according to the length information to obtain a shot slice set iR; (4) sorting according to the sequences of the shot slices in the iR in video data; and (5) merging all the shot slices in the set iR into a plurality of candidate fragments according to a candidate fragment merging algorithm.
Description
Technical field
The invention belongs to video query method technical field, particularly relate to a kind of video information section querying method.
Background technology
In recent years, the research about film clips retrieval is less, and existing research is document. In document, the location of film clips is divided into two steps, first passes through single frames contrast and obtains all representing the frame that frame is similar with film clips, and the combined frames and the video stream file that then recycle segment video contrast, by Continuity Analysis realization being accurately positioned film clips. The method has the disadvantage in that owing to the video data volume is huge, and single frames reduced time complexity is high, and premise is that each frame in film clips is more similar. Along with the multiformity that film clips make, each frame of film clips has been likely to bigger difference; Secondly by segment video positioning stage, arranging of siding-to-siding block length can affect the final degree of accuracy positioned.
Summary of the invention
The present invention is aiming at the problems referred to above, it is provided that a kind of quickly video information section querying method.
For achieving the above object, the present invention comprises the following steps.
(1) shot segmentation is carried out with string matching algorithm.
(2) each camera lens isochronous surface blending algorithm is fused to camera lens section iq, extracts its low-level image feature.
(3) circulation: for each camera lens section iq, retrieves the k similar to it similar camera lens section in camera lens slice library, and filters out incoherent camera lens according to length information; Obtain camera lens set of slices iR.
(4) sort according to camera lens section order in video data in iR.
(5) all of camera lens section in set iR is merged algorithm according to candidate segment and merge into several candidate segment.
(6) it is fused to section according to each candidate segment ic start frame in video library and end frame, calculates the comprehensively degree of likeness in form of candidate segment section and query fragment according to its sight factor, size distribution factor etc.
(7) similar fragments of result set is exported according to sequencing of similarity.
As a kind of preferred version, of the present invention 3) circulation: each camera lens is cut into slices iq, retrieves the k similar to it similar camera lens in camera lens slice library and cut into slices, set a threshold value T, retrieve all Similar sections more than T.
As another kind of preferred version, the present invention initializes ic for sky, each camera lens in R is cut into slices, if it meets time continuity constraints, is then fused in ic; It is empty for otherwise initializing ic+1, continues to merge next camera lens section.
Beneficial effect of the present invention.
The film clips search method that the present invention proposes is divided into three phases: first stage is shot segmentation, then filters out incoherent camera lens according to the visual signature of film clips to be retrieved and length information; Second stage carries out camera lens slice-merging and obtains the candidate segment of Query Result; Candidate segment is fused to isochronous surface by three phases, is then finally retrieved result according to film clips section its low-level image feature of feature extraction. It is demonstrated experimentally that the method has good performance.
The present invention is based on Spatial-temporal slice, and first the visual information according to the length information of news film clips and camera lens section carries out coarse filtration, is then accurately positioned again, all achieves good effect from retrieval rate and effect.
Detailed description of the invention
The present invention comprises the following steps.
(1) shot segmentation is carried out with string matching algorithm.
(2) each camera lens isochronous surface blending algorithm is fused to camera lens section iq, extracts its low-level image feature.
(3) circulation: for each camera lens section iq, retrieves the k similar to it similar camera lens section in camera lens slice library, and filters out incoherent camera lens according to length information; Obtain camera lens set of slices iR.
(4) sort according to camera lens section order in video data in iR.
(5) all of camera lens section in set iR is merged algorithm according to candidate segment and merge into several candidate segment.
(6) it is fused to section according to each candidate segment ic start frame in video library and end frame, calculates the comprehensively degree of likeness in form of candidate segment section and query fragment according to its sight factor, size distribution factor etc.
(7) similar fragments of result set is exported according to sequencing of similarity.
Described 3) circulation: for each camera lens section iq, retrieves the k similar to it similar camera lens section in camera lens slice library, sets a threshold value T, retrieve all Similar sections more than T.
The present invention initializes ic for sky, each camera lens in R is cut into slices, if it meets time continuity constraints, is then fused in ic; It is empty for otherwise initializing ic+1, continues to merge next camera lens section.
Because television station staff is everlasting, some special-effects are done at the beginning and end place of film clips, the film clips of same program are likely in length and to start ending place slightly different, sectioning image has comparatively distinct textural characteristics simultaneously, and thus the present invention adopts the method that dominant color of block mediation textural characteristics combines to retrieve. The color characteristic of isochronous surface adopts divided group dominant hue.
The present invention is set to the 110% of the totalframes of inquiry film clips, reason is in that for film clips, assume that certain two camera lens is cut into slices in same candidate segment, and the interval time between them is more than this threshold value, so there is missing inspection camera lens and missing inspection camera lens totalframes more than query fragment between the section of the two camera lens, now the two camera lens section is hardly possible in same candidate segment, and hypothesis test, so it is more reasonable that threshold value is set to inquiry film clips totalframes. User can according to design conditions sets itself.
Key parameter and the different characteristic impact on retrieval effectiveness in experimental analysis searching algorithm, and contrast with existing document. The accuracy rate of experimental result of the present invention refers to position that film clips the position degrees of offset relative to physical location, and degrees of offset is more big, and accuracy rate is more low. Whether recall ratio principal measure retrieves all similar film clips.
Above content is the further description present invention made in conjunction with concrete preferred implementation; it cannot be assumed that specific embodiment of the invention is confined to these explanations; for general technical staff of the technical field of the invention; without departing from the inventive concept of the premise; some simple deduction or replace can also be made, all should be considered as belonging to the protection domain that the submitted claims of the present invention are determined.
Claims (3)
1.A kind of video information section querying method, it is characterised in that comprise the following steps:
(1) shot segmentation is carried out with string matching algorithm;
(2) each camera lens isochronous surface blending algorithm is fused to camera lens section iq, extracts its low-level image feature;
(3) circulation: for each camera lens section iq, retrieves the k similar to it similar camera lens section in camera lens slice library, and filters out incoherent camera lens according to length information; Obtain camera lens set of slices iR;
(4) sort according to camera lens section order in video data in iR;
(5) all of camera lens section in set iR is merged algorithm according to candidate segment and merge into several candidate segment;
(6) it is fused to section according to each candidate segment ic start frame in video library and end frame, calculates the comprehensively degree of likeness in form of candidate segment section and query fragment according to its sight factor, size distribution factor etc.;
(7) similar fragments of result set is exported according to sequencing of similarity.
2.A kind of video information section querying method according to claim 1, it is characterized in that described 3) circulation: for each camera lens section iq, camera lens slice library is retrieved the k similar to it similar camera lens section, sets a threshold value T, retrieve all Similar sections more than T.
3.A kind of video information section querying method according to claim 2, it is characterised in that it is empty for initializing ic, each camera lens in R is cut into slices, if it meets time continuity constraints, is then fused in ic; It is empty for otherwise initializing ic+1, continues to merge next camera lens section.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410622906.7A CN105630833A (en) | 2014-11-08 | 2014-11-08 | Video information slice query method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410622906.7A CN105630833A (en) | 2014-11-08 | 2014-11-08 | Video information slice query method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN105630833A true CN105630833A (en) | 2016-06-01 |
Family
ID=56045778
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410622906.7A Pending CN105630833A (en) | 2014-11-08 | 2014-11-08 | Video information slice query method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105630833A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113079420A (en) * | 2020-01-03 | 2021-07-06 | 北京三星通信技术研究有限公司 | Video generation method and device, electronic equipment and computer readable storage medium |
| CN115630191A (en) * | 2022-12-22 | 2023-01-20 | 成都纵横自动化技术股份有限公司 | Time-space data set retrieval method and device based on full-dynamic video and storage medium |
-
2014
- 2014-11-08 CN CN201410622906.7A patent/CN105630833A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113079420A (en) * | 2020-01-03 | 2021-07-06 | 北京三星通信技术研究有限公司 | Video generation method and device, electronic equipment and computer readable storage medium |
| CN115630191A (en) * | 2022-12-22 | 2023-01-20 | 成都纵横自动化技术股份有限公司 | Time-space data set retrieval method and device based on full-dynamic video and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2020147857A1 (en) | Method and system for extracting, storing and retrieving mass video features | |
| EP2661709B1 (en) | Method and apparatus for comparing videos | |
| CN104133875B (en) | Face-based video labeling method and face-based video retrieving method | |
| CN104504109A (en) | Image search method and device | |
| CN107180056B (en) | Method and device for matching segments in video | |
| CN103617263A (en) | Automatic TV advertisement movie clip detection method based on multi-mode features | |
| CN102663398A (en) | Color image color feature extraction method and device thereof | |
| JP2014506366A (en) | Method and apparatus for comparing pictures | |
| US20120163475A1 (en) | Fast matching system for digital video | |
| CN103605666A (en) | Video copying detection method for advertisement detection | |
| CN105630833A (en) | Video information slice query method | |
| CN105320710A (en) | Illumination variation resistant vehicle retrieval method and device | |
| CN104991953A (en) | Coarse and fine granularity video searching method based on reverse index | |
| CN103902569A (en) | Video matching method based on Bag of Words | |
| US10394888B2 (en) | Video search system and method | |
| CN105631428B (en) | Video comparison identification method and device | |
| Yang et al. | Multiscale video sequence matching for near-duplicate detection and retrieval | |
| Zhao et al. | Matching logos for slow motion replay detection in broadcast sports video | |
| CN101673267B (en) | Method for searching audio and video content | |
| CN102663359A (en) | Method and system for pedestrian retrieval based on internet of things | |
| CN104199950A (en) | Method of searching for academic papers on basis of fast matching of image similarities | |
| CN104794210A (en) | Image retrieval method combining visual saliency and phrases | |
| Memon et al. | Region based localized matching image retrieval system using color-size features for image retrieval | |
| CN104636330A (en) | Related video rapid searching method based on structural data | |
| CN105472407A (en) | Automatic video index and alignment method based on continuous image features |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20160601 |
|
| WD01 | Invention patent application deemed withdrawn after publication |