Detection of Rice Grain Chalkiness Level with Volume Estimation from Image Processing

Grain chalkiness is an unpleasant trait adversely affecting appearance and milling quality of rice. It causes from packing of carbohydrate non-uniformly in the grain due to fluctuation of rain, humidity and heat. Chalkiness affect the quality and the price of rice directly. To categorize the rice grain chalkiness, the effective tool has not yet been implemented to be used widely in Thailand. Mostly, human detection has been carried out to classify the grain quality. Here we present the FTIR result of grain chalkiness from rice sample obtained from Department of Rice, Ministry of Agriculture of Thailand. Rice grain with chalkiness level of 1 to 5 were investigated. From FTIR result, all samples show the peaks for O-H, C=O and C-H bonds. We found no significant difference in the FTIR peak of all 5 levels of chalkiness which indicates that the cause of loose packing of carbohydrate in chalky area does not originate from microscopic level. UV-Vis spectroscopy showed the significant difference in absorption between chalky and non-chalky area in the visible range around 500-530 nm. Thus, Image processing has been carried out with green light illumination to classify level of rice chalkiness automatically in the aim to replace human detection. Two cameras were set to give perpendicular views of rice grain. Cross sectional thin slab was calculated from image analysis of perpendicular views and was integrated to obtain chalkiness volume and total rice grain volume. Thus, chalkiness level can be acquired. The algorithm was further developed as an application implementing on a mobile phone for practical use in the rice field.