A Computer Based Image Processing Approach to Identify Rice Blast

Abstract

Fungus and bacteria are the main cause of rice plant diseases. Among all fungal diseases rice blast is considered as one of the most common and fatal rice plant disease. Without proper care and use of pesticides this deadly plant disease can cause huge damage for rice crops. Detection of rice blast disease at the early stage can help farmers to use proper pesticides and can save their crops and hence a computerized approach is necessary. Currently a good number of approaches available but none of them seems to provide a suitable solution in terms of identification accuracy. A suitable approach has been presented in this study where both input and output images are color images. Various image processing steps were considered in this study which includes enhancement, noise reduction, color image segmentation, and color features for identification. CNN classifier was applied for validation purpose. In compare to existing available approaches this study proposed approach is capable of providing better results in terms of accuracy which is 97.50%.