Application to Estimate the Rice Polishing Degree Using Image Processing

Abstract

Rice quality evaluation is essential, and a polishing degree is one of the leading quality components contained in the Indonesian National Standard (SNI) requirement on rice. Several methods have been developed to determine the polishing degree, but it is generally subjective and depends on personnel. This study aimed to create a quick detection tool for estimating the polishing degree of rice based on image processing technology which is then developed using the Android system. This rapid detection device was designed to replace the role of an expert in rice physical quality testing with machine learning. The research materials used were long and round varieties of grain. The grain was milled and grounded into rice with 80, 90, 95, and 100 percent polishing degrees. The results of this study indicated that the linear interpolation equation model obtained to estimate the degree of the shape of long-shaped rice was y = 8.52x - 1772.1, and for round rice, it was y = 6.84x - 1404.8, where x was characteristic of the color image, and y was the estimated polishing degree. Based on the result of validation test of the grinding grade using test device, it corresponded with the laboratory test results. This estimator application has been successfully developed as a user-level polishing degree detection tool.