Abstract-Based on a combination of BP neural network and SVM, aluminum plate surface defects classification was discussed. In order to detect the defects, the target image is binaried by adaptive threshold method. After binarizing the target image, extract the characteristic value of six kinds of aluminum plate surface defect images and formed twentyfour dimensional feature vector. The principle and algorithm of BP neural network and support vector machine are introduced, given a way about combination of BP neural network and SVM, and about the important parameters optimization was carried out. The results verify the efficiency, accuracy and robustness of the algorithm about the BP neural network and support vector machine (SVM) classification of combining.