Potential Application of Machine Vision System in Rice Variety Identification
Bernabe L. Paita1
ABSTRACT
A machine vision system was developed to identify rice varieties. Samples of rice or paddy were scanned and analyzed using image processing techniques in order to extract geometric grain features from the sample images. Two-dimensional projections in terms of area or profile and perimeter or edges of the images were two important features extracted to formulate bases for identification and/or classification of the sample rice varieties. Results showed that using circularity as a feature indexer enabled identification of sample rice varieties with accuracy ranging from 61%-92%. A more robust classifier is needed for one rice variety, namely IR72 with a range of circularity value spanning the range of circularity values of three other rice varieties.
1Assistant Professor, Agricultural Machinery Division, Institute of Agricultural Engineering, College of Engineering and Agro-Industrial Technology, University of the Philippines Los Baños, 4031 College, Laguna, Philippines.