Image Segmentation Of Tobacco Brown Spot

Tobacco is an important economic crop in China,which plays an important role in the national economy.However,tobacco brown spot is one of the most harmful diseases to tobacco production in the world.The drug for the prevention and treatment of disease has certain effect,but is caused by the drugs will increase the smoking of tobacco harm people,causing pollution to the environment.Timely harvest of tobacco leaves can effectively avoid the harm of tobacco brown spot.Based on the digital image processing technology,the tobacco leaf can be harvested in time and the yield and quality of tobacco can be improved.Based on the segmentation of tobacco brown spot,it is helpful to improve the accuracy of disease identification.Based on the present research situation at home and abroad,this paper aims to improve the recognition accuracy of tobacco brown spot.The studies on the tobacco brown spot were carried out as follows:(1)To analyze the advantages and disadvantages of Otsu(maximum between-class variance),local threshold method,maximum entropy and iterative method in segmentation of tobacco brown spot.(2)Analyze the problems of edge detection Roberts,Sobel,Canny and LoG operator in the segmentation of tobacco brown spot and improve the LoG operator.(3)Segmentation brown spot tobacco image by K-means method,obtain the lesion color map.(4)The image is segmented by the method of morphology based on K-means image.Concentric wheel for lesion with Canny edge detection operator.(5)The difference of high class feature,classification and recognition of partial two binary tree support vector machines classification accuracy than one-versus-rest support vector machine is about 6%.