Rice quality classification using an automatic grain quality inspection system

In this article, we examine the performance of an automatic inspection system for rice quality classification. Sorting of rice into sound, cracked, chalky, immature, dead, broken, damaged, and off–type kernels was performed by the system. Specific rice quality inspection software was developed to prepare sorting parameters and to refine sorting precision and machine operation. A range–selection algorithm was implemented as a series of parameter range tables. The software was developed in the Windows environment to provide a graphical and user–friendly interface. Results show that the automated inspection system could correctly categorize over 90% of rice kernels based on comparison with human inspection. Results for sound, chalky, and cracked kernels indicated high accuracy in each quality category, around 95%, 92%, and 87%, respectively. The average processing speed for online rice quality inspection was over 1200 kernels/min. This prototype grain quality inspection system demonstrated performance comparable to subjective human inspection.