IIn order to identify the disease by using computer, digital image processing technique and pattern recognition technique were used in the paper, which was composed of image acquisition, image preprocessing, feature extraction and matching. The image is converted from RGB model to HIS. A gray level co-occurrence matrix was used to capture the texture features. The particle swarm optimization algorithm was used to improve the bp neural network algorithm. A maize leaf disease features’ extraction and recognition system was constructed at last. The rationality and validity were validated experimentally.