Abstract: In order to solve the problem of little color difference between different types of Chinese prickly ash after maturity. The image of Chinese prickly ash in blue background was collected, and an automatic extraction algorithm was proposed to extract the shape features of the material. The Chinese prickly ash data set was established and the sample data set was classified based on SVM and DTree classifier. The classification effect of feature plotter was verified, and the influence of different classification algorithms on the classification effect of prickly ash samples was analyzed. The experimental results show that: 1) the classification accuracy of SVM and DTree model based on shape feature descriptors' roundness, rectangularity, fine length and diagonal length reaches 95.87% and 92.12% respectively; 2) when corns are good category, the classification effect of SVM is better, and when seeds are good category, the classification effect of DTree is better.