In this research, a fuzzy inference system (FIS) coupled with image processing technique
was developed as a decision-support system for qualitative grading of milled rice. Two
quality indices, namely degree of milling (DOM) and percentage of broken kernels (PBK)
were first graded by rice processing experts into five classes. Then, images of the same
samples were captured using a machine vision system. The information obtained from
the sample image processing was transferred to FIS. The FIS classifier consisted of two
input linguistic variables, namely, DOM and PBK, and one output variable (Quality), all
in the form of triangle membership functions. Altogether, 25 rules were considered in
the FIS rule base using the AND operator and Mamdani inference system. In order to evaluate
the developed system, statistical performance of the FIS classifier was compared with
the experts’ judgments. Results of analysis showed a 89.8% agreement between the grading
results obtained from the developed system and those determined by the experts.