Categorization and Grading of Grains using Probabilistic Neural Networks

In the contemporary scenario the grading of grain type and quality are recognized manually by visual inspection which is tedious, cumbersome and not accurate. Therefore there is need for the fast and precise system for ascertaining the quality food grains. To overcome this problem an automated system is developed which is used for grain type categorization and identifying quality of rice as 1 grade, 2 grade, and 3 grade can be achieved by using PNN. This proposed system utilizes color and geometrical features as primary attributes for categorization. The grading of food grains sample is based on the size of the grain kernel and presence of impurities. A good categorization accuracy is accomplished using mean of RGB colors and 3 geometrical features. The efficiency of the proposed system for type identification is 98% and the efficiency for quality analysis and grading of rice is 88% and 90% respectively.