CN113920410A - Method for realizing portrait clustering based on multi-data fusion analysis - Google Patents
Method for realizing portrait clustering based on multi-data fusion analysis Download PDFInfo
- Publication number
- CN113920410A CN113920410A CN202111174164.2A CN202111174164A CN113920410A CN 113920410 A CN113920410 A CN 113920410A CN 202111174164 A CN202111174164 A CN 202111174164A CN 113920410 A CN113920410 A CN 113920410A
- Authority
- CN
- China
- Prior art keywords
- data
- face
- human body
- clustering
- analysis
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/253—Fusion techniques of extracted features
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a method for realizing portrait clustering based on multi-data fusion analysis, which comprises the following steps of accessing required video stream data, human face and human body snapshot picture data through an access data module; respectively calling face video structuralization and picture structuralization calculation engines to carry out face and human body structuralization on video stream data and human body snapshot picture data; calling a face clustering analysis algorithm, and performing aggregation association on the same faces through fusion analysis of the place, the time, the structural characteristic value and the structural attribute to finish primary portrait clustering analysis; the video identity ID is created and given through a video identity ID management module; and calling a human body correlation analysis algorithm to complete the correlation analysis of the suspected human body data. On the basis of comparing and analyzing the human face characteristic value data, the invention carries out the analysis of the space-time relationship by fusing the basic information of the equipment and the structural attribute data, thereby greatly improving the precision of data clustering and reducing the aggregation divergence.
Description
Technical Field
The invention relates to the technical field of computers, in particular to a method for realizing portrait clustering based on multi-data fusion analysis.
Background
The technical scheme of face clustering in the prior art mainly comprises the following two implementation modes:
1. and based on a built static face library feature library, carrying out real-time feature value comparison on the accessed face snapshot data, and directly hanging the snapshot pictures meeting the set threshold under related personnel. The disadvantages of this solution are: firstly, the existing static face library is excessively dependent, and when static data support cannot be provided, the clustering effect is extremely common; the second is fusion analysis without considering other data.
2. The method aims at a cluster analysis algorithm of face snapshot data, but only aims at similarity comparison of face structural characteristic values, and carries out cluster combination on snapshot pictures with standard similarity. The disadvantages of this solution are: with the rapid growth of the access data volume and the archival cluster cardinality, archival cluster errors can be amplified, clustering accuracy is reduced, and divergence is increased.
The common problems existing in the two technical schemes are as follows: firstly, introducing more auxiliary analysis data is not considered so as to improve the clustering precision; secondly, correlation aggregation is not carried out on the face-related human body snapshot data; thirdly, the accuracy of filing and clustering is poor in large-range and mass data environment.
Disclosure of Invention
The invention aims to overcome the defects in the prior art and provides a method for realizing portrait clustering based on multi-data fusion analysis.
The technical scheme of the invention is as follows:
the method for realizing portrait clustering based on multi-data fusion analysis comprises the following steps:
the required video stream data, the human face and human body snapshot picture data are accessed through the access data module;
respectively calling face video structuralization and picture structuralization calculation engines to carry out face and body structuralization on video stream data and human body snapshot picture data so as to output and obtain corresponding face and body structuralization data;
basic information data of all face snapshot devices are docked;
calling a face clustering analysis algorithm, performing aggregation association on the same faces through fusion analysis of the location, time, structural characteristic values and structural attributes, outputting face clustering data, and completing one-time portrait clustering analysis;
the video identity ID is created and given through a video identity ID management module;
calling a human body correlation analysis algorithm to complete correlation analysis of suspected human body data;
and storing and writing the portrait clustering analysis structure data into a database.
The human face cluster analysis algorithm comprises the following steps:
accessing processed human face and human body structural data, including characteristic values and structural attribute data;
loading basic information data of the accessed face snapshot equipment;
loading the existing full-scale clustering face data;
comparing the face characteristic values, and carrying out face structural attribute association analysis based on snapshot time;
performing space-time relation analysis based on the snapshot time and place of the associated equipment, and removing interference data from the primary clustering data;
and finishing the target clustering data updating.
The human body correlation analysis algorithm comprises the following steps:
accessing human body structural data to be subjected to correlation analysis, wherein the human body structural data comprises characteristic values and structural attribute data;
loading basic information data of the accessed face snapshot equipment;
based on the clustered face data, finding out strongly related human body data through the face;
analyzing the human body snapshot image data, and finding out related human face data through the human body;
performing space-time relation analysis based on the snapshot time and place of the associated equipment to remove interfering human body data;
and finishing the correlation updating of the suspected human body data of the target cluster.
Furthermore, the cover photo of each portrait target clustering file can be automatically optimized and updated according to the quality of the picture or the latest snapshot.
The invention has the beneficial effects that:
1. compared with the prior art, the method does not need to rely on various face static libraries;
2. on the basis of the comparison and analysis of the human face characteristic value data, the spatio-temporal relation analysis is carried out by fusing the basic information and the structured attribute data of the equipment, so that the data clustering precision is greatly improved, and the aggregation divergence is reduced;
3. the invention can disassemble a large-area large system into a small-area small system, thereby reducing filing clustering errors caused by the precision problem of the image identification technology;
4. according to the invention, the clustering relation of the suspected human body snapshot data is increased, so that the retrieval and analysis efficiency of suspected targets is effectively improved, and the GPU computing resource consumption is reduced.
Drawings
FIG. 1 is a flow chart of a method for implementing portrait clustering based on multi-data fusion analysis;
FIG. 2 is a flow diagram of a face cluster analysis module;
FIG. 3 is a flow chart of a human body correlation analysis module.
Detailed Description
For a better understanding of the invention, reference will now be made to the following examples and accompanying drawings.
The method for realizing portrait clustering based on multi-data fusion analysis comprises the following steps:
s01: the required video stream data, the human face and human body snapshot picture data are accessed through the access data module;
s02: respectively calling face video structuralization and picture structuralization calculation engines to carry out face and body structuralization on video stream data and human body snapshot picture data so as to output and obtain corresponding face and body structuralization data;
s03: basic information data of all face snapshot devices are docked;
s04: calling a face clustering analysis algorithm, performing aggregation association on the same faces through fusion analysis of the location, time, structural characteristic values and structural attributes, outputting face clustering data, and completing one-time portrait clustering analysis;
s05: the video identity ID is created and given through a video identity ID management module;
s06: calling a human body correlation analysis algorithm to complete correlation analysis of suspected human body data;
s07: and storing and writing the portrait clustering analysis structure data into a database.
Specifically, the face clustering analysis algorithm in step S04 includes the following steps:
s11: accessing processed human face and human body structural data, including characteristic values and structural attribute data;
s12: loading basic information data of the accessed face snapshot equipment;
s13: loading the existing full-scale clustering face data;
s14: comparing the face characteristic values, and carrying out face structural attribute association analysis based on snapshot time;
s15: performing space-time relation analysis based on the snapshot time and place of the associated equipment, and removing interference data from the primary clustering data;
s16: and finishing the target clustering data updating.
Specifically, the human body correlation analysis algorithm in step S06 includes the following steps:
s21: accessing human body structural data to be subjected to correlation analysis, wherein the human body structural data comprises characteristic values and structural attribute data;
s22: loading basic information data of the accessed face snapshot equipment;
s23: based on the clustered face data, finding out strongly related human body data through the face;
s24: analyzing the human body snapshot image data, and finding out related human face data through the human body;
s25: performing space-time relation analysis based on the snapshot time and place of the associated equipment to remove interfering human body data;
s26: and finishing the correlation updating of the suspected human body data of the target cluster.
Preferably, step S06' after step S06 may automatically optimize updating the cover photograph of each portrait target cluster file according to picture quality or recent snapshot.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, so any modifications, equivalents or improvements made within the spirit and principles of the present invention should be included in the scope of the present invention.
Claims (4)
1. The method for realizing portrait clustering based on multi-data fusion analysis comprises the following steps:
the required video stream data, the human face and human body snapshot picture data are accessed through the access data module;
respectively calling face video structuralization and picture structuralization calculation engines to carry out face and body structuralization on video stream data and human body snapshot picture data so as to output and obtain corresponding face and body structuralization data;
basic information data of all face snapshot devices are docked;
calling a face clustering analysis algorithm, performing aggregation association on the same faces through fusion analysis of the location, time, structural characteristic values and structural attributes, outputting face clustering data, and completing one-time portrait clustering analysis;
the video identity ID is created and given through a video identity ID management module;
calling a human body correlation analysis algorithm to complete correlation analysis of suspected human body data;
and storing and writing the portrait clustering analysis structure data into a database.
2. The method for clustering human images based on multiple data fusion analysis according to claim 1, wherein: the steps of the face clustering analysis algorithm are as follows:
accessing processed human face and human body structural data, including characteristic values and structural attribute data;
loading basic information data of the accessed face snapshot equipment;
loading the existing full-scale clustering face data;
comparing the face characteristic values, and carrying out face structural attribute association analysis based on snapshot time;
performing space-time relation analysis based on the snapshot time and place of the associated equipment, and removing interference data from the primary clustering data;
and finishing the target clustering data updating.
3. The method for clustering human images based on multiple data fusion analysis according to claim 1, wherein: the human body correlation analysis algorithm comprises the following steps:
accessing human body structural data to be subjected to correlation analysis, wherein the human body structural data comprises characteristic values and structural attribute data;
loading basic information data of the accessed face snapshot equipment;
based on the clustered face data, finding out strongly related human body data through the face;
analyzing the human body snapshot image data, and finding out related human face data through the human body;
performing space-time relation analysis based on the snapshot time and place of the associated equipment to remove interfering human body data;
and finishing the correlation updating of the suspected human body data of the target cluster.
4. The method for clustering human images based on multiple data fusion analysis according to claim 1, wherein: and the cover photo of each portrait target clustering file can be automatically optimized and updated according to the picture quality or the latest snapshot.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111174164.2A CN113920410A (en) | 2021-10-08 | 2021-10-08 | Method for realizing portrait clustering based on multi-data fusion analysis |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202111174164.2A CN113920410A (en) | 2021-10-08 | 2021-10-08 | Method for realizing portrait clustering based on multi-data fusion analysis |
Publications (1)
Publication Number | Publication Date |
---|---|
CN113920410A true CN113920410A (en) | 2022-01-11 |
Family
ID=79238426
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202111174164.2A Pending CN113920410A (en) | 2021-10-08 | 2021-10-08 | Method for realizing portrait clustering based on multi-data fusion analysis |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN113920410A (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114333039A (en) * | 2022-03-03 | 2022-04-12 | 济南博观智能科技有限公司 | Portrait clustering method, device and medium |
CN114639143A (en) * | 2022-03-07 | 2022-06-17 | 北京百度网讯科技有限公司 | Portrait filing method, equipment and storage medium based on artificial intelligence |
CN117931738A (en) * | 2024-03-21 | 2024-04-26 | 南京启数智能***有限公司 | Portrait file track treatment method and system based on road network reachability |
-
2021
- 2021-10-08 CN CN202111174164.2A patent/CN113920410A/en active Pending
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN114333039A (en) * | 2022-03-03 | 2022-04-12 | 济南博观智能科技有限公司 | Portrait clustering method, device and medium |
CN114639143A (en) * | 2022-03-07 | 2022-06-17 | 北京百度网讯科技有限公司 | Portrait filing method, equipment and storage medium based on artificial intelligence |
CN114639143B (en) * | 2022-03-07 | 2024-04-16 | 北京百度网讯科技有限公司 | Portrait archiving method, device and storage medium based on artificial intelligence |
CN117931738A (en) * | 2024-03-21 | 2024-04-26 | 南京启数智能***有限公司 | Portrait file track treatment method and system based on road network reachability |
CN117931738B (en) * | 2024-03-21 | 2024-06-07 | 南京启数智能***有限公司 | Portrait file track treatment method and system based on road network reachability |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113920410A (en) | Method for realizing portrait clustering based on multi-data fusion analysis | |
US20130124796A1 (en) | Storage method and apparatus which are based on data content identification | |
US20140244595A1 (en) | Context-aware tagging for augmented reality environments | |
CN111324665B (en) | Log playback method and device | |
US20200019474A1 (en) | Consistency recovery method for seamless database duplication | |
WO2020087950A1 (en) | Database updating method and device, electronic device, and computer storage medium | |
CN112395157B (en) | Audit log acquisition method and device, computer equipment and storage medium | |
CN113901279B (en) | Graph database retrieval method and device | |
CN111444274A (en) | Data synchronization method, data synchronization system, and apparatus, medium, and system thereof | |
CN111723161A (en) | Data processing method, device and equipment | |
TW202211081A (en) | Target data updating method, electronic equipment and computer readable storage medium | |
CN111079653B (en) | Automatic database separation method and device | |
WO2018019310A1 (en) | Big data system data backup and recovery methods and devices, and computer storage medium | |
CN111694853A (en) | Lineage-based data increment acquisition method and device, storage medium and electronic equipment | |
CN111061916A (en) | Video sharing system based on multi-target library image recognition | |
JP2021520016A (en) | Database update method and equipment, electronic devices, and computer storage media | |
WO2022252322A1 (en) | Method for synchronizing memory database and relational database of feature tag-based power grid monitoring system | |
US11880405B2 (en) | Method for searching similar images in an image database using global values of a similarity measure for discarding partitions of the image database | |
CN112307251B (en) | Self-adaptive recognition correlation system and method for knowledge point atlas of English vocabulary | |
CN110297842B (en) | Data comparison method, device, terminal and storage medium | |
CN110362703B (en) | Picture arranging method and system | |
US11822797B1 (en) | Object computational storage system, data processing method, client and storage medium | |
US10649979B1 (en) | System, method, and computer program for maintaining consistency between a NoSQL database and non-transactional content associated with one or more files | |
CN114969041B (en) | Multi-source main and auxiliary entity identity discrimination and data self-supplementing processing method | |
CN112749748B (en) | Data increment clustering method, system, medium and device based on reduction |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |