Using artificial intelligence system to determine Rials-value of white rice grains based on appearance properties and percentage of chalky grains

Qualitative grading and classification of rice are one of the factors affecting the appearance quality and customer satisfaction. The objective of this study was to develop an accurate algorithm for diagnosing three important qualitative indices of rice including percentage of broken grains, percentage of chalky grains and degree of milling. Some rice samples of a variety namely 'AliKazemi' were provided from Rice Research Institute of Iran. After questioning expert persons, three quantitative levels (Low, Medium and High) were defined. Each of the 270 samples of rice grains were photographed under identical conditions. Afterwards, image quality improvement, rice image segmentation and textural, color and size features identification were performed using Matlab Software and image processing methods. In development of fuzzy inference model for the product pricing, the Mamdani inference system was applied and defuzzification process was done using Center of Area defuzzifier. Results indicated that the average accuracy of the image processing algorithm in detecting percentage of broken grains, percentage of chalky grains and degree of milling was respectively equal to 95.2%, 90.74% and 98.99%. The decisions made by the fuzzy model were consistent with the expert’s judgment in 95.2% of the cases. This technic can be good alternative to manually which takes less time and more accurately assess the characteristics of the product key.