Universal open type face identification system

Abstract

The invention relates to a universal open type face identification system which abstracts a related data in a face identification algorithm into a database data by introducing a backstage database support, adopts modularization and layering design and uses the backstage database as a bridge to perform the functions of collecting face samples, managing a face identification data, training a face classifier, guiding in a standard face database, testing for static picture face identification, testing for dynamic real-time face identification, generating a testing report, etc. Each module not only can work independently but also can work with other modules to finish tasks such as testing, real-time identification, etc. The invention can perform different face identification algorithms under an open type universal frame system, can finish face identification tasks in different databases and under different conditions, and can provide a reference model for the other face identification system.

Description

technical field [0001] The invention relates to a general open face recognition system, which introduces the support of the database, takes generality, openness and distribution as the design goal, and completes various face recognition tasks independently or cooperatively through each functional module. The design of the face recognition system provides a reference prototype. It is mainly used in the design of face recognition system and the research field of face recognition algorithm. Background technique [0002] After about 40 years of development in face recognition research, researchers have launched various fields such as face detection and face tracking, detection and calibration of face feature points, face feature description and feature extraction, and face feature classification. Extensive and in-depth research, various theories, and various algorithms emerge in endlessly. At the same time, the development technology of practical systems has reached a certain m...

AI Technical Summary

Problems solved by technology

[0005] The Face Identification Evaluation System was developed by Colorado State University in the United States, which provides four basic face recognition algorithms (Eigenfaces, Combination PCA and LDA, Bayesian Intrapersonal / Extrapersonal Image Difference Classier, Elastic Bunch Graph Matching), and can obtain standard face databases The test results of FERET (Face Recognition Technology) on the above four algorithms, but the system uses text to save the experimental data, the reading and writing efficiency is relatively low, and the scalability of the system is very poor, it is difficult for users to embed their own face recognition algorithms carry out testing

FRGC (Face Recognition Grand Challenge), developed under the auspices of NIST (National Institute of Standards and Technology), is a face recognition algorithm evaluation system commonly used in academia (Overview of the Face Recognition Grand Challenge, IEEEConference on Computer Vision and Pattern Recognition 2005), the system is mainly aimed at testing face recognition algorithms under high-resolution face images, 3D faces and face image preprocessing. Users can easily and quickly embed their own face recognition algorithms into the evaluation system for testing. Experimental data It is stored in XML file format, which speeds up the reading and writing efficiency, but the system only supports FRGC face database and does not support other standard face databases, and the above system can only test image-based face recognition algorithms, not video-based face recognition algorithms. Based face recognition algorithm test

Images