CN103186772A - Face recognition system and method based on cluster framework - Google Patents
Face recognition system and method based on cluster framework Download PDFInfo
- Publication number
- CN103186772A CN103186772A CN2013100361560A CN201310036156A CN103186772A CN 103186772 A CN103186772 A CN 103186772A CN 2013100361560 A CN2013100361560 A CN 2013100361560A CN 201310036156 A CN201310036156 A CN 201310036156A CN 103186772 A CN103186772 A CN 103186772A
- Authority
- CN
- China
- Prior art keywords
- module
- image
- face
- control module
- information
- 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
Abstract
The invention discloses a face recognition system and method based on a cluster framework. The face recognition system comprises a front end collecting module, a face characteristic extracting module, an index control module, a plurality of matching search modules, an image service control module and a plurality of image file management modules. For a large number of databases, data searching and matching capability can be effectively improved and higher data throughput can be provided by adopting the face recognition system and method based on the cluster framework, so that operating efficiency of a background is greatly improved.
Description
Technical field
The present invention relates to people's face and detect identification and distributed search technical field, particularly face identification system and the method under a kind of cluster framework.
Background technology
Along with socioeconomic fast development, social population's flowing velocity is accelerated year by year, recognition of face as an important technical of identification to relating to public safety, business administration, the importance of every field such as social management highlights gradually.In recent years, along with academia's progressively going deep into a series of subject studies of recognition of face, proposed multiple effective theoretical foundation and practice scheme, the method for main flow has in the world at present: the method for masterplate coupling, learn-by-example, neural network and opportunity hidden Markov model.Although face recognition technology has wide application prospect, no matter be at discrimination, still on antifalsification, all with fingerprint, the retina lamp has bigger gap.It is low to present recall precision, and average delay is big, the low inferior problem of processing power under the big data.Particularly be applied to customs, the city rail is handed over, railway, and under the occasion that civil aviaton etc. are densely populated or flow of the people is bigger, face identification system is difficult to long-time stable operation, is faced with the bottleneck of load-bearing capacity.
Summary of the invention
At the defective that exists in the above-mentioned prior art, the objective of the invention is: propose face identification system and method under a kind of cluster framework, it simplifies the backstage search procedure by adopting distributed computing technology and hashing algorithm, the search procedure of parallelization simultaneously, finally provide high the data throughput capabilities of total system, having solved existing face identification system can't be in the long-time problem of using of the bigger occasion of data acquisition amount.
The technical solution adopted for the present invention to solve the technical problems is: the face identification system under the cluster framework, and described face identification system comprises:
Be used for the front end collection module that front end image acquisition, pre-service and effective information extract;
According to the image information that the front end collection module provides, be used for the facial feature extraction module of face-image feature selecting, extraction;
According to the face characteristic image information that the facial feature extraction module provides, be used for the index control module of search and control search procedure;
Some match search modules for the Search Results that constitutes described index control module;
The Search Results that provides according to the index control module is used for the managing image system and calls the images serve control module of image file; And
Some be used to finishing described image file, and the image file management module to the terminal user is provided.
As preferred version, described front end collection module further comprises:
Image capture module is used for collection and the storage of front end image;
Image modification module is used for the correction image content and extracts eye locating information and deflection information;
The image pretreatment module; Be used for the location and grasp human face image information, and be divided into some index zone; And
Image output module; According to the some index zone that is divided into, be used for generation and output and meet retrieval needs facial image collection.
As preferred version, described facial feature extraction module further comprises:
ASM and AAM algorithm call submodule, are used for the performance advantage according to people's face exterior contour and inner major organs, make up combination performance model, and foundation combination performance model is to each sample extraction proper vector;
The PCA algorithm calls submodule, is used for realizing hiding non-notable feature, and reduces the dimension of proper vector;
Low-dimensional data set generation module; Be used for according to described ASM with the AAM algorithm calls submodule and described PCA algorithm calls the result that submodule finally obtains, generation comprises the low dimension data collection of some proper vectors.
As preferred version, some described match search modules are managed in the parallel control of described index control module.
As preferred version, some described image file management modules are managed in the parallel control of described images serve control module.
Face identification method under the cluster framework may further comprise the steps:
Step S1: gather human face image information, and carry out the content correction and extract eye locating information and deflection information;
Step S2: according to the eye locating information that provides among the step S1 and deflection information, face region, people from location, and the human face region after will locating is divided into some zones, forms a facial image collection;
Step S3: according to the performance advantage of people's face exterior contour and inner major organs, call ASM and AAM algorithm and make up combination performance model, and according to making up the performance model to each sample extraction proper vector;
Step S4: call the dimension that the PCA algorithm reduces described proper vector, and generate a low dimension data collection;
Step S5: utilize the index control module that described low dimension data collection is called the match search module in batches one by one this locality storage data are searched for, and provide rreturn value;
Step S6: described index control module is collected described rreturn value, and the screening of occuring simultaneously, and The selection result is generated a Search Results;
Step S7: utilize images serve control module calling graph to finish the required image file as document management module, merge and go to heavy back to submit the terminal user to.
As preferred version, described step S5 comprises the steps: also that specifically described match search module uses the hash table of local Hash matches function retrieval self maintained, and the special image data structure of extracting in the corresponding data bucket is returned.
The invention has the beneficial effects as follows: adopt the present invention when facing the database of magnanimity, effectively expedited data is searched for matching capacity, its parallelization search and image management ability can provide higher data throughout for system, and mass data is cut apart storage, it had both guaranteed the integrality of data, improve simultaneously throughput and the data redudancy of system again, accelerated the retrieval reading speed of image, promoted running background efficient significantly.
Description of drawings
Fig. 1 is system architecture synoptic diagram of the present invention;
Fig. 2 is the frame structure figure of front end collection module;
Fig. 3 is the frame structure figure of facial feature extraction module;
Fig. 4 is the invention process method flow diagram.
Embodiment
In conjunction with the accompanying drawings, the present invention is further detailed explanation.
As shown in Figure 1, the face identification system under the cluster framework comprises: front end collection module 1, facial feature extraction module 2, index control module 3, some match search modules 4, images serve control module 5; And some image file management modules 6.
The front end collection module 1 that shown in Figure 2 is among Fig. 1, described front end collection module 1 further comprises:
Image pretreatment module 13; Be used for the location and grasp human face image information, and be divided into some index zone; And
Facial feature extraction module 2 among Fig. 1 shown in Figure 3, described facial feature extraction module 2 further comprises:
ASM and AAM algorithm call submodule 21, are used for the performance advantage according to people's face exterior contour and inner major organs, make up combination performance model, and foundation combination performance model is to each sample extraction proper vector;
The PCA algorithm calls submodule 22, is used for realizing hiding non-notable feature, and reduces the dimension of proper vector;
Low-dimensional data set generation module 23; Be used for according to described ASM with the AAM algorithm calls submodule and described PCA algorithm calls the result that submodule finally obtains, generation comprises the low dimension data collection of some proper vectors.
Its course of work of face identification system among the present invention is as follows: the front end collection module is gathered image, and correction image content and extraction eye locating information and deflection information, then, the image pretreatment module is separated original image set and rough handling, these zones are also cut apart in face region, people from location, form one group of facial image collection that meets the retrieval needs; The facial feature extraction module is used ASM and AAM algorithm to finish the eigenwert of individual human face image is extracted, and recycling PCA algorithm reduces dimension, forms the low-dimensional data acquisition that comprises some vectors; The index control module is called the match search module in batches one by one and is stored data search at this locality according to data acquisition, and the match search module is used in the hash table of local Hash matches function retrieval self maintained, and extracts the special image data structure in the corresponding data bucket and return; The index control module is collected rreturn value, utilizes the screening of occuring simultaneously, and obtains the result that satisfies condition and constitutes Search Results; The images serve control module is according to the front Search Results, and calling graph is finished the image file that requires as document management module, merges to go to heavy back to submit the terminal user to.
Its concrete implementing procedure is as shown in Figure 4: the face identification method under the cluster framework may further comprise the steps:
Step S1: gather human face image information, and carry out the content correction and extract eye locating information and deflection information;
Step S2: according to the eye locating information that provides among the step S1 and deflection information, face region, people from location, and the human face region after will locating is divided into some zones, forms a facial image collection;
Step S3: according to the performance advantage of people's face exterior contour and inner major organs, call ASM and AAM algorithm and make up combination performance model, and according to making up the performance model to each sample extraction proper vector;
Step S4: call the dimension that the PCA algorithm reduces described proper vector, and generate a low dimension data collection;
Step S5: utilize the index control module that described low dimension data collection is called the match search module in batches one by one this locality storage data are searched for, and provide rreturn value;
Step S6: described index control module is collected described rreturn value, and the screening of occuring simultaneously, and The selection result is generated a Search Results;
Step S7: utilize images serve control module calling graph to finish the required image file as document management module, merge and go to heavy back to submit the terminal user to.
Described step S5 comprises the steps: also that specifically described match search module uses the hash table of local Hash matches function retrieval self maintained, and the special image data structure of extracting in the corresponding data bucket is returned.
More than show principal character and the innovative point of only having described this programme.Those skilled in the art should understand, and this programme is not restricted to the described embodiments.Under the prerequisite that does not break away from the innovation point and protection domain, this programme also has various variations, and these changes and improvements all will fall in the claimed scope of this programme.The claimed scope of this programme is limited by appending claims and equivalent thereof.
Claims (7)
1. the face identification system under the cluster framework, it is characterized in that: described face identification system comprises:
Be used for the front end collection module that front end image acquisition, pre-service and effective information extract;
According to the image information that the front end collection module provides, be used for the facial feature extraction module of face-image feature selecting, extraction;
According to the face characteristic image information that the facial feature extraction module provides, be used for the index control module of search and control search procedure;
Some match search modules for the Search Results that constitutes described index control module;
The Search Results that provides according to the index control module is used for the managing image system and calls the images serve control module of image file; And
Some be used to finishing described image file, and the image file management module to the terminal user is provided.
2. face identification system as claimed in claim 1, it is characterized in that: described front end collection module further comprises:
Image capture module is used for collection and the storage of front end image;
Image modification module is used for the correction image content and extracts eye locating information and deflection information;
The image pretreatment module; Be used for the location and grasp human face image information, and be divided into some index zone; And
Image output module; According to the some index zone that is divided into, be used for generation and output and meet retrieval needs facial image collection.
3. face identification system as claimed in claim 1, it is characterized in that: described facial feature extraction module further comprises:
ASM and AAM algorithm call submodule, are used for the performance advantage according to people's face exterior contour and inner major organs, make up combination performance model, and foundation combination performance model is to each sample extraction proper vector;
The PCA algorithm calls submodule, is used for realizing hiding non-notable feature, and reduces the dimension of proper vector;
Low-dimensional data set generation module; Be used for according to described ASM with the AAM algorithm calls submodule and described PCA algorithm calls the result that submodule finally obtains, generation comprises the low dimension data collection of some proper vectors.
4. face identification system as claimed in claim 1, it is characterized in that: some described match search modules are managed in the parallel control of described index control module.
5. face identification system as claimed in claim 1, it is characterized in that: some described image file management modules are managed in the parallel control of described images serve control module.
6. the face identification method under the cluster framework is characterized in that: may further comprise the steps:
Step S1: gather human face image information, and carry out the content correction and extract eye locating information and deflection information;
Step S2: according to the eye locating information that provides among the step S1 and deflection information, face region, people from location, and the human face region after will locating is divided into some zones, forms a facial image collection;
Step S3: according to the performance advantage of people's face exterior contour and inner major organs, call ASM and AAM algorithm and make up combination performance model, and according to making up the performance model to each sample extraction proper vector;
Step S4: call the dimension that the PCA algorithm reduces described proper vector, and generate a low dimension data collection;
Step S5: utilize the index control module that described low dimension data collection is called the match search module in batches one by one this locality storage data are searched for, and provide rreturn value;
Step S6: described index control module is collected described rreturn value, and the screening of occuring simultaneously, and The selection result is generated a Search Results;
Step S7: utilize images serve control module calling graph to finish the required image file as document management module, merge and go to heavy back to submit the terminal user to.
7. the face identification method under a kind of cluster framework as claimed in claim 6, it is characterized in that: described step S5 comprises the steps: also that specifically described match search module uses the hash table of local Hash matches function retrieval self maintained, and the special image data structure of extracting in the corresponding data bucket is returned.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2013100361560A CN103186772A (en) | 2013-01-30 | 2013-01-30 | Face recognition system and method based on cluster framework |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2013100361560A CN103186772A (en) | 2013-01-30 | 2013-01-30 | Face recognition system and method based on cluster framework |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103186772A true CN103186772A (en) | 2013-07-03 |
Family
ID=48677933
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2013100361560A Pending CN103186772A (en) | 2013-01-30 | 2013-01-30 | Face recognition system and method based on cluster framework |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103186772A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108900793A (en) * | 2018-07-25 | 2018-11-27 | 武汉恩智电子科技有限公司 | A kind of recognition of face video playback system based on video monitoring |
| CN108984614A (en) * | 2018-06-12 | 2018-12-11 | 成都三零凯天通信实业有限公司 | A kind of visible image method for quickly identifying under the environment based on big data |
| CN109583264A (en) * | 2017-09-28 | 2019-04-05 | 阿里巴巴集团控股有限公司 | Information identifying method, device and electronic equipment |
| CN109635663A (en) * | 2018-11-14 | 2019-04-16 | 南宁学院 | A kind of Distributive System of Face Recognition |
| CN109795942A (en) * | 2019-01-17 | 2019-05-24 | 杭州海康睿和物联网技术有限公司 | Staircase control system, staircase monitoring device and its intelligent control method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080123967A1 (en) * | 2006-11-08 | 2008-05-29 | Cryptometrics, Inc. | System and method for parallel image processing |
| CN101447021A (en) * | 2008-12-30 | 2009-06-03 | 爱德威软件开发（上海）有限公司 | Face fast recognition system and recognition method thereof |
| US20100202703A1 (en) * | 2009-02-09 | 2010-08-12 | Sungkyungwan University Foundation For Corporate Collaboration | Real-time face detection apparatus |
| CN102201061A (en) * | 2011-06-24 | 2011-09-28 | 常州锐驰电子科技有限公司 | Intelligent safety monitoring system and method based on multilevel filtering face recognition |
-
2013
- 2013-01-30 CN CN2013100361560A patent/CN103186772A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080123967A1 (en) * | 2006-11-08 | 2008-05-29 | Cryptometrics, Inc. | System and method for parallel image processing |
| CN101447021A (en) * | 2008-12-30 | 2009-06-03 | 爱德威软件开发（上海）有限公司 | Face fast recognition system and recognition method thereof |
| US20100202703A1 (en) * | 2009-02-09 | 2010-08-12 | Sungkyungwan University Foundation For Corporate Collaboration | Real-time face detection apparatus |
| CN102201061A (en) * | 2011-06-24 | 2011-09-28 | 常州锐驰电子科技有限公司 | Intelligent safety monitoring system and method based on multilevel filtering face recognition |
Non-Patent Citations (2)
| Title |
|---|
| 孙拔群: "社交网络中的多媒体数据挖掘", 《中国优秀硕士学位论文全文数据库信息科技辑》, no. 5, 15 May 2012 (2012-05-15), pages 35 - 45 * |
| 李鹏: "人脸识别中的若干算法研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》, no. 7, 15 July 2010 (2010-07-15) * |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109583264A (en) * | 2017-09-28 | 2019-04-05 | 阿里巴巴集团控股有限公司 | Information identifying method, device and electronic equipment |
| CN108984614A (en) * | 2018-06-12 | 2018-12-11 | 成都三零凯天通信实业有限公司 | A kind of visible image method for quickly identifying under the environment based on big data |
| CN108900793A (en) * | 2018-07-25 | 2018-11-27 | 武汉恩智电子科技有限公司 | A kind of recognition of face video playback system based on video monitoring |
| CN109635663A (en) * | 2018-11-14 | 2019-04-16 | 南宁学院 | A kind of Distributive System of Face Recognition |
| CN109795942A (en) * | 2019-01-17 | 2019-05-24 | 杭州海康睿和物联网技术有限公司 | Staircase control system, staircase monitoring device and its intelligent control method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102364498B (en) | Multi-label-based image recognition method | |
| CN103678670B (en) | Micro-blog hot word and hot topic mining system and method | |
| CN103593371B (en) | Recommend the method and apparatus of search keyword | |
| CN103235825A (en) | Method used for designing large-quantity face recognition search engine and based on Hadoop cloud computing frame | |
| CN103186772A (en) | Face recognition system and method based on cluster framework | |
| JP2015512095A (en) | Method, apparatus and computer readable recording medium for image management in an image database | |
| CN105046720B (en) | The behavior dividing method represented based on human body motion capture data character string | |
| CN109710792A (en) | A kind of fast face searching system application based on index | |
| CN109800416A (en) | A kind of power equipment title recognition methods | |
| CN103366120A (en) | Bug attack graph generation method based on script | |
| CN104615734A (en) | Community management service big data processing system and processing method thereof | |
| CN110019070A (en) | A kind of security log clustering method based on Hadoop and system of calling to account | |
| CN105678244A (en) | Approximate video retrieval method based on improvement of editing distance | |
| CN109241315B (en) | Rapid face retrieval method based on deep learning | |
| Yin | Clustering microtext streams for event identification | |
| CN107301203B (en) | Mass data comparison method and system | |
| CN103020630B (en) | The disposal route of characteristics of image and device | |
| US11681753B2 (en) | Geotagged video spatial indexing method based on temporal information | |
| CN111625596A (en) | Multi-source data synchronous sharing method and system for real-time consumption scheduling of new energy | |
| CN110110598A (en) | The pedestrian of a kind of view-based access control model feature and space-time restriction recognition methods and system again | |
| US20170004148A1 (en) | Visual search method, system and mobile terminal | |
| CN105512155A (en) | Device and method for multi-layer semantic image retrieval | |
| CN104199950A (en) | Method of searching for academic papers on basis of fast matching of image similarities | |
| CN111768066B (en) | Park electric heating load coupling relation analysis method and device based on fusion characteristics | |
| CN111783891B (en) | Customized object detection method |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20130703 |