An Intelligent Firmness Sensor for An Automated Herring Roe Grader

Abstract

ABSTRACTThis article presents a knowledge-based approach for firmness assessment of herring roe, for application in an automated grading machine. A low-level ultrasound sensor along with knowledge-based preprocessing and high-level decision making is interpreted as an intelligent firmness sensor, in the present context. Techniques of low-level sensing and information preprocessing are presented. The rule-based decision making system for quality assessment uses fuzzy logic concepts, which incorporates the expertise of experienced human graders. The membership functions are estimated for firmness resolution, using a probabilistic approach. It is found that this approach is superior to the use of trapezoidal membership functions.