Based Face Recognition Password System Development And Research

As part of the object recognition problem, deeply research on face detection and recognition has become an urgent demand. It has been an attractive task and also has sufficient practical value.With the rapid development of the society, the rapid and efficient status validation has been needed under many circumstances. As the internal property of a person, the biologic feature is steady and has obvious individual differences. So it's the perfect criterion for status validation. Using the face feature for status validation is a natural and directly method. Compare with other biologic features, face feature is more convenient and friendly to use, which the users will accept easily. Different from features like iris and fingerprint, the feature of human face can be easily collected without the cooperation of persons, which has great meaning for antiterrorism. At the same time, face detection and recognition also has extensive practical value in the research field like digital video processing.With regard to the face detection, an Adaboost based algorithm is integrated with the system, which achieves real time face detection function. This thesis also makes extensive discussion on the superiority and difficulties of the LDA based face recognition algorithm. Moreover, we propose a novel method to calculate the discriminant vectors, which can be applied directly on high-dimensional image data and avoid the computational difficulties. With the implementation of the new algorithm, the detection rate has been increased. Result of the experiment on face image database shows that our algorithm has superiority to the related face recognition algorithm of present time.The innovative work of this thesis is mainly on the research of face recognition algorithm and improvement on LDA based algorithm. A new method to compute the discriminant vectors is provided. Besides the research, I focus the engineering work on the implementation of the video data collection and real time face detection and recognition.