Classification Algorithm Based On Fuzzy Genetic Optimization Of Support Vector-based Human Face

Face recognition is a difficulty in the research on pattern recognition and machine vision. In the first place, face recognition algorithms are studied in this paper. Accordingly, face classification algorithm based on fuzzy genetic algorithms (FGAs) and support vector machines (SVMs) are proposed. In the next place, in order to fulfill the demand of safety-check, checking on work attendance, identity, and so on, a face recognition system is developed independently in this paper.Firstly, the background on the development of face recognition technologies and the purpose of the paper are introduced completely in this paper; and then, based on correlative literatures, face database, face feature extraction and face classification technologies are expatiated, so that some innovations of this paper can be educed.The paper contains two innovations as follows:Firstly, SVMs have been applied in face recognition, but in practice, the problem on how to select parameters of SVMs is not solved properly, so that its application is restricted. In order to get the optimal parameters of SVMs automatically, a parameter selection approach based on FGAs is proposed in this paper. The crossover probability and mutation probability of genetic algorithms (GAs) are adjusted on-line based on fuzzy logic. In addition, the dimension of feature vector has an influence on classification result, so the dimension is also adjusted by FGAs.Secondly, a novel face recognition system is designed and developed based on some technologies containing C++, Microsoft Foundation Class (MFC), DirectShow, Oracle, and so on. The system contains six modules as follows:1. The module of video capture: The module is mainly compiled by DirectShow. It's used to capture video to picture.2. The module of face detection: Adaboost algorithm is adopted by the module.3. The module of face track: The next face position will be detected in the neighbor region of the last face position.4. The module of face feature extraction: Principle Component Analysis (PCA) is adopted by this module.5. The module of face classification: In view of recognition speed, Neural Network with impulse is adopted by this module.6. The module database management: The module is mainly compiled by Oracle.On account of classification rate of face recognition, face classification based on FGAs and SVMs have cheerful prospect; otherwise, face recognition system will be applied well in the scopes including passageway management, identity, social safety, intelligent surveillance, man-machine interaction.