Healthy leaves Identification using Multi SVM Approach

Abstract

The leaves of the plant can identify health of the plant. Leaves plays a very important role in identification of plant disease, whichis beneficial for the farmers to prevent losses, and helps in increasing the quality and productivity of agricultural product. It is very difficult to identify plant disease manually and itrequires a lot of time. Thus, automatic identification is becoming popular day by day. The proposed method presents an image processing based technique based on health of leaves that helps in quickly recognizing the health of plant. This paper gives improved plant disease detection and classification method using Multi SVM approach.It can detect six types of diseases i.e. Anthracnose, Bacterial Blight, black spot,Alternaria, Cercospora and Healthy leaves.The user can know the affected area of leaf in percentage by identifying the disease properly.The user can rectify the problem in very efficient manner. Disease detection stepsfollowed by the paper are pre-processing, segmentation, feature extraction and classification. K-means clustering algorithm is usedin recognizing the defected area. The proposed methodhave used a Gray-Level co-occurrence matrix (GLCM) for obtaining the statistical measures from the matrix of a given image. The proposed method used Multi-Class Support Vector Machine (MSVM) for classification. The MSVM approach is more efficient, since the feature vector may contain different features that helps in easily identifying the health of leaves.
Keywords: Segmentation, Plant, K-means, Multi SVM