Research On Face Recognition Algorithm Based On Geometric Feature

Nowadays, with the rapid development of electronic technology and biological information technology, The pattern recognition technique which is based on the biological characteristics obtain the widespread attention. Face recognition technology is the rapid development of the current biometric identification technology and it is widely applied. Face recognition technology is a very challenging research topic in the multidisciplinary and multi-field, the main areas are: image processing and analysis, pattern recognition and machine learning, artificial neural networks and mathematical disciplines, such as artificial intelligence and machine vision. Face recognition process includes three stages :face image preprocessing, feature extraction and pattern classification. This paper mainly studies face recognition technology which is based on the geometric feature, also the theory and process of face recognition technology is studied and discussed.This paper introduces the research background and significance of face recognition, scope of application, research status at home and abroad and the it is commonly used in the process of recognition of human face database and self-built database of faces. For face recognition process, we have studied the key technologies, and discussed the problems appeared in the process, the analysis summarizes research process in the light of difficulties and problems.we mainly introduced the use of face recognition algorithm based on geometric features in this paper. Face image preprocessing is presented in this paper, we used the image enhancement, image smoothing, edge detection and extraction and image normalized operation, after various operations of rendering is presented. In facial organs location and feature extraction, we used the edge detection and projection function of face, eyes, nose and mouth and other major organs to locate, and extract the main organs of 12 feature points. Using 12 feature point 13 eigenvector is constructed, the 13 structure feature vector is not affected by image displacement and rotation changes, it has the very good stability. To feature vector for the normalized operation, it make major organs in face facial reflect its distribution very good. Ratio in the classification of face recognition process, we used the weighted function which can identify the weighted ratio function to calculate each eigenvalue of the difference in the characteristic vector, considering the characteristic value of different importance for identifying the results of different influence, the function and the weighted ratio is the use of feature vector variance characteristics to calculate the weight, calculation of the method is simple, calculation of the weights of high accuracy, it can improve the recognition rate and suitable for the identification process.Identification process of the experiment operation is in the face of self-built test in library and ORL face database. Experimental results show that the face recognition algorithm based on geometric features conveniently, calculating the volume of data is small, and the recognition rate is high, the identification effect is ideal.