The Study of Face Recognition Based on Radial Basis Function Neural Networks

• Main Authors: Chien-Pin Wu, 吳建斌
• Other Authors: Rei-Yao Wu
• Format: Others
• Language: zh-TW
• Published: 2007
• Online Access: http://ndltd.ncl.edu.tw/handle/54792520231890524431

碩士 === 世新大學 === 資訊管理學研究所(含碩專班) === 95 === The thesis build a complete face detection and face recognition system by support vector machine and radial basis function neural network. The system detects possible human face in normalized RGB color space. The detected skin color blocks are send to the support vector machine for face detection. In order to obtain a correct region, the process of noise reduction, connected component labeling, component adjustment, image normalized, wavelet transformation are applied before the support vector machine is performed. The detected human face is then sent to a radial basis function neural network for recognition. Experiments on 256 test pictures that contain 383 faces show that the contracted system performs face detection in the rate of 92.2%. The result of face recognition experiment is also great. Face recognitions for the ORL face database, the simple background pictures, and the complex background pictures achieve the recognition rate of 100%, 85% and 75% respectively.