Plant Disease Image Processing System

In order to use pesticide rationally and effectively to prevent plant disease, the farmers must get the plant growing information exactly and rapidly. Curing the disease depends on the pathogen and the degree of plant disease.Firstly, a set including hand-carrier and digital camera was built. Secondly, using cucumber leaf disease speckle information, the method of improving vector median filtering to preprocessing was carried out for cucumber image and its result was compared with Ideal Filter's . As the result shows, the method increases the operation speed and the image quality. Based on faintness and indeterminacy of Cucumber Downy Mildew disease, this research adopted statistical pattern recognition of Fisher linear distinguish method to segment image successfully under RGB color space. Based on this, utilizing open and close operation of mathematical morphology, clearing up the alone spot, stab and filling small empty hole is realized as to plant disease image after segment. Image preprocessing effect is ideal. As a result, segmentation effective percent was above 97.8% after validation. Finally, 5 boundary features were picked up as inputs of the plant disease diagnostic model. After obtained the massive data sample carries on the nerve network the training .3 layers united Bp neural network was built to diagnose the disease including 5 features as network inputs and the hidden cell number was 30. The net error was satisfied depending on demand after training.This study investigated technologies of plant disease image processing and diagnosis using digital processing, neural network and statistical pattern recognition method based on arithmetic theory and VC++ language.