Abstract
Agriculture sector has been facing extensive challenges such as climate instability, cropping pattern, inadequate use of fertilizers, disease identification, and promoting new technologies. Owing to this, recognizing plant disease is one of the leading concerns in boosting productivity. In this paper, an efficient image processing technique has been utilized to classify diseases that occur in rice plant. Initially, background portion of an RGB rice plant image is removed in preprocessing phase. Next, three different clusters are obtained from the image through K-means clustering algorithm. Later, the diseased portions from these clusters of image are retrieved using histogram and color values. Color and texture features are calculated on diseased images. Finally, these obtained features are subjected to the classification phase using a support vector machine (SVM) classifier. Experimentation validates that the proposed detection method through SVM achieves maximal accuracy of 83.3% by outperforming other existing methods.
