Research And Application On The Face Recognition Technology

The technology of human face recognition is an active subject in the field of pattern recognition. There are broad application in the fields of laws, business and safety systems ect. For the particularity of human face images, face recognition with a computer is a very difficult problem and there are still many works to do before such technology can be used wildly. With the development of the society, the application of face recognition systems will be wilder and brings much challenge to the researchers.The human face recognition system is a kind of pattern recognition system based on Information processing. It can be divided into three parts: face detection, feature extraction and pattern classification. The first part is to detect the face through the input image or video. The second part is to find out a set of features that can represent the images from different persons; The third part classifies the features got from the second part. The performance of the system depends on both of the three parts.For the first part of the system, the method used in the thesis is the haar-like face detection which is the most novel algorithms nowadays. It can provide with high accurate and meet the request for real time detection. The method is based on haar-like features,cascade classifier and the Adaboost arithmetic. It can detect face at any scale at any position, and suit for muli-face detection. The algorithm was realized in computer with the help of OpenCV, during the experiment, the reorientation problem was found, after some amelioration of the Algorithm, the problem was solved.For the second part of the system, the Independent component analysis was used here. Compare with PCA and other method for feature extraction, the ICA use the high rank statistic information effectively, and can improve the rate of recognition.For the third part of the system, the minimax probability machine was introduced to the system, and was realized in the computer. Compare to the most famous classification method such as SVM and NN, the great value of MPM is that it can provide an explicit upper boundary of misclassification for the future data and give us an intuitionistic factor for system evaluation.