Segmentation Method And Application For Wheat Diseases Image Under Complex Background

As one of main grain crops of China, wheat is an important guarantee of food security for our country. Wheat disease is one of the key factors that can influence the wheat yield, and the outbreak of the disease can lead to a massive reduction of output and quality. Therefore, to strengthen the study of wheat disease recognition method is of great importance and value. Among all the methods, the image recognition technology is a new technology applied in the field of wheat disease identification and it has been rapidly developed and widely used in recent years.Image recognition includes image preprocessing, image segmentation, feature extraction classification and recognition and other steps. Among them, the image segmentation is the key for image recognition, which determines the subsequent identification effect. Image segmentation under a complex environment has long been a difficult problem to solve in the image segmentation field. Due to the complex environment (light shadow, soil and weeds, etc.) of wheat field, and the huge amount of background information, the traditional segmentation method is not enough to segregate their image of diseased spot from the complex background environment, and some of the present proposed algorithms haven’t been well solved.On the basis of summarizing the relevant research results at home and abroad, this paper aims at the practical characteristics of wheat plant disease image under the complex background has taken four kinds of leaf diseases of wheat (stripe rust, brown leaf rust, powdery mildew and leaf blight) as the study object, studies the image segmentation method of wheat diseases under complex background, and has also verified the effectiveness of the proposed method through the experiment, and has eventually designed the intelligent image segmentation system for wheat disease under the complex background. This thesis mainly studies this subject from the following aspects:(1)This thesis has put forward a segmentation method which is based on combination ofK-means clustering method and various other algorithms, and the method adopts the step by step segmentation strategy to segregate the image of diseased spot from the complex background other surroundings:firstly, in the L*a*b* color space through making use of the differences of soil, weeds and other background and the diseased wheat leaf presenting in the a*b* component, and to take K-means clustering algorithm for pixels in a*b* color component and at last segregate the image of wheat leaf from the complex background; Secondly, using the method of Otsu dynamic threshold to start binarization processing, and combining with the mathematical morphology operation and the area threshold method to segregate main leaf images of diseased wheat with disease spots; Finally, according to the differences of color feature between on the healthy leaf areas and on the disease spots of disease spots and to adopt super green splitting method so as to realize the image segmentation of disease spot of wheat under the complex background.(2) The verification experiment of the method of segmentation. When using the above methods to take segregation experiment for the colored images of 4 kinds of common wheat diseases under the complex background, and according to the comparison of the experimental results and statistical analysis of experimental data, it is concluded that accuracy of the algorithm of the segmentation in more than 95%, showing that the algorithm is of great robustness, high segmentation accuracy which has laid a foundation for the recognition of diseases.(3) It has designed and developed the intelligent image segmentation system based on MATLAB language for wheat disease under the complex background.