Face Recognition System Design And Implementation Based On Principal Component Analysis

Biometric technique, which makes use of physical and behavior characteristics of human for personal identification, has got more attention out of many personal identification methods, for its higher security, reliability and effectiveness. Human face recognition is a biometric technique that is easily accepted. It has important research and application values and has been an active research field in recent years.In this paper, we study and implement a face-recognition based personal authentication system using Modified PCA. This paper consists of three parts as follow.Firstly, the paper describes face preprocessing in detail which is an important step in the face recognition. The face preprocessing methods we adopt are based on image processing techniques. The main purpose is to get the standardized facial images, and to eliminate the impact of illumination to some extent. In this paper, several key preprocessing methods are introduced, such as geometry normalization, gray-scale normalization and images binary-conversion.Secondly, we compare four face recognition algorithms based on PCA. We introduce the classical vector-based images of principal component analysis (PCA) in details. In order to overcome some deficiencies of the classical PCA, we propose a modified PCA (M-PCA). And then we introduce a two-dimensional principal component analysis (2DPCA) based on a direct matrix. Meanwhile, a modified 2DPCA algorithm (M-2DPCA) is proposed. The study found that the recognition rate of M-2DPCA is close to 2DPCA, and higher than the recognition rate of M-PCA. However, we find that M-2DPCA and 2DPCA methods have higher space complexity and time complexity in the process of reconstruction. Therefore they will occupy more memory in the process of reconstruction. When the actual application of the system could not provide enough memory, M-PCA exhibits best performance.Finally, we implement face recognition system based on M-PCA, including the system framework, the regedit module and the logon module in face recognition system and some key technologies used in the system. We choose the M-PCA method when implementing the system, because M-PCA method can not only increase the recognition rate, but effectively accelerate the speed of the recognition.A full personal identification system is designed in this paper, which combines algorithm optimization and system design. The system takes many factors into account, such as image requisition devices, scene setting, etc. to boost the algorithm efficiency.