Research And Implement Of Face Recognize Based On HMM

Face recognition is a challenging task in the fields of pattern recognition, image processing, computer vision and so on. However, it is a hot issue recently. Broad face recognition includes two aspects of face detection and face identification. This paper describes the principle and process of implementation based on hidden Markov model. The main research is following.1. Using the unique biological characteristics of face, to complete the design and implementation of face detection algorithm based on Haar-Like Features.2. To extract observation vector using DCT (Discrete Cosine Transform), and to implement the face training and recognition, the observation vector will be used in face recognition algorithm based on HMM (Hidden Markov Model). A design and algorithm implementation based on two-dimensional HMM is suggested, and applied to the process of face recognition to improve the accuracy.3. The basic architecture of OpenCV and its advantage to develop face recognition systems is explained, and its application in the face recognition is detailed.4. Applying the designed face recognition algorithm to software system, and the overall program design of face recognition as well as the implementation of each functional module are illuminated in detail.5. Through many experiments, face recognition result is tested and analyzed on the designed software systems.In this paper, the design can be applied to video surveillance, security access control, identification and other security fields. It can also be used for video games, smart cameras and the other fields. In a word, it has broad application prospects.