Texture analysis for diagnosing paddy disease

The objective of this research is to develop a diagnosis system to recognize the paddy diseases, which are Blast Disease (BD), Brown-Spot Disease (BSD), and Narrow Brown-Spot Disease (NBSD). This paper concentrates on extracting paddy features through off-line image. The methodology involves converting the RGB images into a binary image using variable, global and automatic threshold based on Otsu method. A morphological algorithm is used to remove noises by using region filling technique. Then image characteristics consisting of lesion percentage, lesion type, boundary color, spot color, and broken paddy leaf color are extracted from paddy leaf images. Consequently, by employing production rule technique, the paddy diseases are recognized about 87.5 percent of accuracy rates. This prototype has a very great potential to be further improved in the future.