To achieve the computer identification of orange secco plant disease and the scientific evaluation of disease levels,the paper proposed a identification method through the analysis of disease image, automatic extract the effective feature, design classifier model.In the paper method was studied how to enhancement processing the diseases of image, segmentat disease spot, extract feature, and Construct classifier model, etc. Then for example five of confusion between the diseases, extracting the disease spots the tone and texture, shape characteristics, after optimization respectively by using support vector machine (SVM) and BP neural network to identify disease categories. The experimental results show that this method can be a very good recognition plant disease categories for scientific control and give a scientific evaluation for the plant disease harm degree.