Research Of Image Recognition Based On Neural Network

With the development of society, the request to recognize one effectively and automatically has been urgent day by day. Facies , as one of the biological characteristic's attribute, has very strong stability and individual independency. Thus it is the ideal source of information for ID verification. It has more advantage, such as natural and direct, than other biological characters. So, it is more acceptable in our society.The definition of face recognition can be described as follows: Given a certain static picture or dynamic video picture, try to detect and recognize one or more person in it based on the information of pre-built face database. The process of face recognition can be divided into three parts: Face Detection, Feature Extraction and Face Recognition. The main task of this thesis lies in the following aspects:1. Face Detection .In this paper we built a system based on Adaboost algorithm. Result showed us that the system can detect frontal images including a wide range of formats . The experiment with CMU databases showed that the system reached a high hit-rate and low false-alarm-rate.2. Feature Extraction. The algorithm of Eigenface is widely used in pattern recognition based on Principle Component Analysis, which can represent resource date in a lower dimensional space .3. Face Recognition. We have given the advantages and disadvantages of the BP algorithm, which has been widely used in pattern recognition, and introduce moment and learning rate to improve BP algorithm. We also show how to design BP network classifier and how to put it into effect in face recognition. At last, a rapid face detection and recognition system is built based on neural network. The experiments showed us that the hit-rate in ORL database can reach 95.5%.The experiment results show that the system is quite successful and achieves a high hit-rate , low false-alarm-rate and high recognition speed.