Ear Recognition Based On Principal Component Analysis

Along with the high-speed development of computer techniques and information technology, the information safety and public safty become unprecedented important. The methods of personal identification that are both compenient and safe are needed by the people ergently. Under this background, the personal identification techniques that use the characteristics of human body are developed more quickly. As a kind of new identificable technique, ear recognition becomes the hottest point gradually with its special advantages in this area.This paper mainly studies the approaches to the features extraction and the recognition in the ear database. The main contents are as follows:1. Athough the geometrtic method is viewed dircetly, the effects in the pratical ear identification are not very good. Therefore, this paper doesn't use the tradional metod but adops the means of Principal Component Analysis (PCA) based on the K-L transformation.2. In the process of classification by using the BP to identify person's ear, the author puts forward the improved project and optimizes BP, according to the shortages of restraining slowly and strange values produced by BP. With the optimized BP, the author carried on two experiments and analysized the datas of the experiments.3. The result of experiment displayed that the traditional method of Principal Component Analysis (PCA) have the localization. In this paper, GA with its special characteristics of ear pictures is introduced to solve this problem;4. By the usage of the Visual C++ 6.0 released by Microsoft Company, the author designed the software of automatical ear identification system striding towards the practicality, which makes a meaningful try in this area.