Identification and Classification of Foreign Bodies from Rice Grains Using Digital Image Processing

Abstract

This research article presents the method for the identification and classification of foreign bodies from rice grains using the digital image processing approach. Any matter other than the rice grains in the rice images is considered as the foreign bodies in the current research work. The foreign bodies can be in terms of weed, stones, soil lumps, pieces of stems, plant leaves, or other types of grains. The amount of the foreign bodies decides the quality of the rice grains which in turn may be helpful to determine the quality of the rice grains. In the manual inspection of grains, the foreign bodies are evaluated based on the manual inspection which may not be so accurate in every inspection. Whereas the machine vision system automatically determines the amount of the foreign bodies present in the rice grains which can be helpful to farmers in sowing the seeds and also marketing the rice grains. The digital image analysis algorithms are developed using machine learning techniques using MATLAB and Python programming languages to determine the foreign bodies present in the rice grain samples using a neural network method. The foreign bodies in the rice grain image samples are determined based on the color texture and morphological features using a digital image processing method. All three above mentioned features are presented to the neural network for training purpose and this trained network is later used to identify the foreign bodies in rice grains.