A Comparative Study Of The Image Recognition Algorithm For Minorities

In recent years, Face Recognition gradually become a hot research projects.Because Face Recognition was used in the field of information security, identity and other advantagesas.Currently research on Face Recognition is based on the original method of optimization or lead into new algorithms, and often using standard database for experiments, such as the ORL face database and the Yale face database.But there are a few of people to study specific populations face image. Xinjiang is a multi-ethnic region, so for ethnic minorities in Xinjiang face image recognition algorithm has important significance.Firstly I through the D70 S camera to get Xibe, Hui, Kirgiz, Uighur, Kazak filming six ethnic minorities face images, and through post processing to constitute face database. Then I used Principal Components Analysis,PCA+BP neural networks and PCA+Support Vector Machine recognition algorithms to research the homemade ethnic minorities face database. Through the Matlab simulation experiment to compare the advantages and disadvantages of these types of face image recognition algorithms. Through the identification of results can be seen using the PCA+SVM algorithm in recognition rates and the recognition time is better than the other two methods, especially using Radial Basis Function(RBF) kernel PCA+SVM to identify more effective.