Study On Face Recognition Theory And Its Application

Face Recognition Technology (FRT), as the best biometric technology, is becoming a topic research in the computer vision and pattern recognition due to its potential commercial worth. This thesis is concerned with some achievements about FRT. One of author's achievements is that an approach for human face detection with complex background using Genetic Algorithms (GAs) is proposed, and the experimental results are satisfying.In the first part of this thesis, a survey of the different methods about FRT is given based on a large number of literatures.In the second part of this thesis, a new method about face detection using GAs is proposed. Firstly, author proposes a new GA, which can quicken population convergence to optimum due to its capability to sustain diversity in population. Secondly, an efficient presentation of human face pattern is given according to geometric template and gray distribution. Then the proposed GAs is used to optimize the procedure of face detection with complex background in different scales. The experiments demonstrate that our'method achieves high accuracy and the average computation time is less than conventional ones.In the third part of this thesis, the algorithm of our automatic face recognition system is presented in details. In the sub-system efface detection, based on some other scholars' research, face is detected using an analytic-to-holistic approach, which is the combination of various methods. In the sub-system of face recognition, an algorithmfor face recognition using pseudo 2-D Hidden Markov Models is introduced. In the end of this part, the evaluation result of our system is given.