Detection and Classification of Leaf Diseases using Texture and Color Feature

• Main Authors: Yu-Sheng Wang, 王昱勝
• Other Authors: Kuo-Guan Wu
• Format: Others
• Language: en_US
• Published: 2019
• Online Access: http://ndltd.ncl.edu.tw/cgi-bin/gs32/gsweb.cgi/login?o=dnclcdr&s=id=%22107NCHU5441026%22.&searchmode=basic

碩士 === 國立中興大學 === 電機工程學系所 === 107 === Using the leaf characteristics of plants to carry out disease analysis and detection, not only contributes to the development of agricultural automation, but also monitors the growth of plants in an instant, and early detection of pests and diseases, thereby increasing crop yields. Traditionally artificial It is very time-consuming and laborious to carry out crop diseases. Recently, in the research of automatic detection of diseases by plants, it is common practice to first cut out the lesions in the leaves, and the texture color is then passed through an appropriate classifier to distinguish the types of diseases that may be infected. However, the effect of lesion region segmentation will directly affect the accuracy of subsequent disease detection. Especially in the early stage of the disease, the lesion area is small, or it is affected by the background light, which will cause the segment performance of the lesion area to decrease. A classification method of plant diseases based on leaf texture is proposed. Without the segmentation processing of lesions, the texture features and color features obtained by GLCM (Gray Level Co-occurrence Matrix) parameters are directly obtained, and then SVM (Support Vector Machine) classification processing is performed. Achieve nearly 95% classification accuracy.