Corn Leaves Department Disease Spot Recognition

In recent years, along with the computer processing ability and machine vision technology’s continuousdevelopment, digital image processing and image recognition have been developing rapidly. Our country isone large agricultural nation, corn is one of China’s important food crops, the occurrence of disease, insect pest,crop smothering are very serious. It is possible to combine image recognition technology and knowledge ofphytopathology to recognize the disease spot in the corn leaves. This article mainly and conclusions are asfollows:(1) Image acquisition：Learning the relevant phytopathology knowledge of corn leaves diseases, and aplan is drawn up to acquire the images of corn leaves. The related hardware equipments are selected tocomplete corn leaves disease spot image acquisition.(2) Image preprocessing: The image preprocessing is used to the image of diseases spot in corn leaves.Homomorphic filtering and median filtering technology are adopted to process the images of corn leaves, thequality of image improve obviously.(3) Image segmentation: Researching the image segmentation methods that are generally used to segmentthe crop disease spot, after the comparison of two segmentation methods, method based on LXF level set isbetter than method based on fuzzy-C clustering for image segmentation. After that, the morphologic operationis adopted to process the diseases spot in the segmented image, the results are more close to the true diseasesspot.(4) Feature extraction: firstly, the relevant features in diseases spot image of different kinds of diseases areextracted, then the optimization filter of the feature parameters are completed.18kinds of feature parameters(including the texture characteristics, the color characteristics, morphological characteristic) are extracted fromthe image, the12feature parameters among of them that can delegate the feature better are selected afterfiltered by genetic algorithm.(5) Disease spot recognition: learning the basic method of pattern recognition, Bayes classification andfuzzy pattern recognition algorithm are applied to experiment, the average rate of accuracy that adopted fuzzypattern recognition algorithm is above93%.(6) System development: the Visual C++is adopted to be implementation tool, integrating the module ofimage preprocessing, image segmentation, the feature extraction and recognition of diseases spot in cornleaves to implement the system of diseases spot recognition in corn leaves.