Classification of Basmati Rice Grains using Image Processing Techniques

This paper presents a method of classifying five varieties of Basmati rice grains (including a non-Basmati rice) using three different types of features namely, the morphological, color and texture. The classification process has been carried out on two different kinds of images of rice. One type of image where rice grains are scattered and isolated whereas, in other case, bulk rice grain images are taken. From the isolated rice images, only the morphological features were extracted. Again, from the bulk rice images, color and texture features were extracted. Two classification methods, namely, the SVM and BPNN classifiers have been adopted here to classify the rice grains. It is seen that the Color and Texture features give better results compared to morphological features alone. From the experimental results, it can be observed that the maximum accuracy achieved is 96.4% with BPNN classifier.