In the narrow sense, Face Recognition refers to identity confirmation or identity search through the face of a person. At present, face recognition technology has matured and different types of commercial systems have been put into use. The face recognition system establishes an automatic face recognition alarm network to automatically identify the specific person in a specific area, and the system is directional alarm when the person who enters the specific area is not registered and authorized. The current face recognition system device has a large volume, poor mobility, is not easy to carry, and is difficult to popularize and widely use. With the ever-changing communication technology and the increasing popularity and popularity of smartphone recording, it has become possible to design a portable face recognition system that combines CBIR technology with mobile communication technology. It not only has the functions of general face recognition system confirmation and verification, but also makes full use of the advantages of wireless communication. It can be widely used in computer or network security, access control, access control and attendance, card management, public security pursuit, out In the different security fields, such as entry border inspection, airport security inspection, driver's license or passport authentication, the portable portable function is not replaceable by the general face recognition system, so it has a wider application prospect.
This article refers to the address: http://
1 CBIR technology CBIR (Content Based Image Retrieva I) is a content-based image retrieval. It belongs to the research field of image analysis and information processing. It refers to the direct use of image content for image information query. The purpose is to give a query image, According to the content information or the specified query standard, the content is consistent or similarly matched in the image database, and finally the corresponding image that meets the query condition is provided.
1.1 The basic principle of CBIR CBIR generally consists of two parts: image indexing system and image retrieval system. The image indexing system indexes the image file and sets the retrieval target, ie the retrieval point, according to the design requirements, to form an ordered indexing system for matching retrieval. The system provides search access to the underlying visual features of different images, such as color, texture, shape, and object, by design function.
1.2 The main content of CBIR The main content of CBIR is color, texture, shape and object. Color features include image color distribution, mutual relationship and composition; texture refers to image texture structure, direction, combination and symmetry relationship; shape refers to image contour composition, shape, size, etc.; object includes image sub-object relationship, quantity, Properties and rotation, etc.
1.3 Characteristics of CBIR CBIR can extract features and semantics directly from images. The retrieval process is directly connected with semantic extraction, which makes the retrieval process more efficient and adaptable. Similar match is used to replace Exact Match. That is, a similar alignment method is used to obtain a similar image structure, and the convergence is asymptotic until the desired result is obtained; the user can browse by selecting an example or drawing a graphic by himself, and can continuously improve the search formula and refine the retrieval process; Search based on objective attributes (keywords), content-based search, multi-level efficient search based on object-related search and concept search.
1.4 Application System of CBIR Technology In the field of CBIR, after more than ten years of theoretical research, many mature algorithms and some valuable systems have been produced. Low-level image information (such as image color, texture, shape, etc.) is commonly used to implement image content query. For example, the QBIC (QueryBy Image Content) image retrieval system developed by IBM Research Center, the Visual SEEK image query system of Columbia University, the PhotoBook system developed by the Massachusetts Institute of Technology Laboratory, and the MARS system of UIUC University of the United States. In order to further improve the accuracy of the search, the CBIR system uses the similarity algorithm to calculate the similarity between the user submission result and the index database, extracts the information that meets the threshold as the result and outputs it in descending order of similarity, and at the same time. During the retrieval process, the user interacts with the user continuously. The system learns the relevant information of these feedbacks and performs the next round of retrieval again to meet the requirements of the user.
2 Web application example based on CBIR technology The system in the above "CBIR technology-based application system" mainly uses the color, texture, shape, etc. of the image to compare the similarities, most of which are based on the underlying visual features of the image, and are not complete. Web implementation solutions are far from universal use. At present, the full implementation of Web-based CBIR technology application website mainly has www. Like. Com and www. Polarrose. Com.
Www. Like. Com is a commercial clothing shopping website that uses unique technology to query and retrieve clothing images. The user submits photos of clothes, shoes, hats, jewelry, etc., and after matching and searching, the website will return the product name, price and other related information of the clothing. The biggest limitation of the system is that the content of the search can only be clothing, and it is a world-renowned product. What's interesting is that by submitting a photo of a celebrity, you can search for something that a celebrity wears and return a list of similar costumes for a shopping reference.
Www. Polarrose. Com uses its own 2D image 3D model transformation technology to provide free facial image retrieval services. The service is implemented in conjunction with the client's running software and server-side processing functions. The client software is publicly available in the form of computer web browsers Firefox and IE client software. When the user browses the website and displays a still image containing the facial image of the person, the client software generates a small logo on the face of the person. By clicking on the logo, an image with a face similar to the character can be retrieved. If the character's name is already registered in the PolarRose database, the name can also be confirmed. If the selected character has not yet registered a name, the user can also register the name or correct the wrong name. Nicholas, vice president of PolarRose, described its function: "Using our face search technology, you can find the same person in different scenes and photos of lights according to some basic features of the face, as long as you can see His face."
3 System design is based on the popularity of the Internet, the maturity of CBIR technology and the successful implementation of Web implementation solutions. The design idea of ​​CBIR, Internet and wireless communication technology is proposed to realize a new type of mobile face recognition based on mobile phone. The system is shown in Figure 1.
The working principle and face recognition process of the system are as follows: the user enters the character to be confirmed through the camera function of the smartphone itself, performs facial feature shooting, sends the picture to the server as a mobile phone short message, and the server stores it in the database. After the image is subjected to the "content-based" matching process, the image information with a high degree of matching is sent back to the mobile phone user as a short message for confirming the character.
3.1 System Operation Process Firstly, the human face image database supported by CBIR technology is established. In the face image database, the facial features are extracted and indexed, and then the facial feature database is stored. This is the basis and premise of the system work. With the support of this foundation and premise, the user sends the image captured by the mobile phone to the server through the wireless network, performs facial feature extraction on the received image, and then performs matching search in the feature database, and uses the image information with high similarity to SMS. The form is sent back to the mobile phone user for identity verification.
3.2 System development tools and operating environment development tools: Microsoft produced and released Visual Studio2008 Tearn Suite system; operating system: Windows 2003 Serv-er; application server: JRun 4.0, JRun is a Java application server developed by Macromedia Provides a fast and reliable J2EE-compatible platform. If you want to add server-side Java functionality to your web application, JRun will be the most sensible and correct choice. Back-end database: Oracle9i, Oracle9i is an Inter-net support relationship object developed by Oracle Corporation. The model's distributed database and highly integrated, intelligent Internet application infrastructure platform are the complete integration of Oracle9i Database, Oracle 9i Application Server and Oracle9i Developer Suite; the number of servers: 2 units.
3.3 System Module Description (1) Database Image Database 1: Use the Spider web spider program to capture the image of the person in the web page and its explanatory text stored in the image database 1 to compensate for the shortage of the number of professional image libraries.
Image Database 2: Consists of professional images and their annotations to provide an authoritative interpretation of the user.
Image Feature Library 1: The image face features in the image database 1 are extracted, and the face code is stored in the image feature library 1.
Image Feature Library 2: Used to store the face code of the image in the image database 2.
"Face Pattern Coding" works according to the essential features and shapes of the face. It resists changes in light, skin tones, facial hair, hair, glasses, expressions and postures. It has strong reliability and can be used from millions of people. Accurately identify a person.
(2) Image feature extractor This module is composed of algorithms. The feature vector method and the face pattern template are the two main algorithms used by the extractor.
The eigenvector method first determines the size, position, distance and other attributes of the facial image such as the iris, nose and mouth angle, and then calculates their geometric feature quantities. These feature quantities form a feature vector describing the face; The template method is to store a number of standard face image templates or face image organ templates in the library. When performing the comparison, the sample face image is matched with all the pixels in the library using normalized correlation metrics. In addition, there is an autocorrelation network using pattern recognition or a combination of features and templates.
(3) Indexer Index is an image reordering module that classifies and indexes image databases and image feature libraries. Indexing the image database is an effective way to optimize the organization structure of the database and improve the efficiency of the system. In fact, it processes the surface code of the image in an orderly manner, which can effectively narrow the search range and improve the response speed of the system.
(4) GUI
The GUI (Graphical User Interface) is the visual experience and interactive operation part of the screen product. The GUI is a human-machine system project that combines computer science, aesthetics, psychology, behavioral science, and analysis of needs in various business fields. It emphasizes the three-person-one environment as a system for overall design. The purpose of this customer-oriented system engineering design is to optimize the performance of the product, make the operation more user-friendly, reduce the cognitive burden of the user, and make it more suitable for the user's operation requirements. The GUI of the system mainly involves the network query part and the mobile phone. section.
(5) Main interface The connection between the mobile phone and the system server through the base station is mainly realized by the CMPP or SGIP protocol. The CMPP and SGIP (ETIP on CD-MA) protocols are respectively addressing the short message Internet access solutions provided by China Mobile and China Unicom. They specify the application layer interface protocol between the information resource station entity and the Internet short message gateway. CMPP and SGIP can provide services for realizing mobile data value-added services, including the following services: Email notification, voicemail notification, Internet short message, mobile platform email, reminder notification, and automatic integrated service information desk.
4 Conclusion The face recognition system described in this paper successfully integrates and applies existing technical achievements such as CBIR technology, Internet technology and mobile phone communication. Its originality lies in the short message service and mobile phone camera function and the Web identification system based on CBIR technology. Organically integrated, this face recognition system based on mobile phone is not only an extension of mobile phone value-added services, but also provides a positive reference and a broader application space for CBIR-based identification systems in many fields.
EVI Heat Pump,Heat Pump System ,Heat Pump Dryer ,Pool Heat Pump
Heat Pump Water Heater Co., Ltd. , http://www.nbheatpumps.com