Deciduous Fruit Grading Robot (Part 2)

Development of Image Processing System

Abstract

An image processing system, which was developed for the grading robot reported in last paper of this series, consisted of 12 color TV cameras and 28 illumination devices. The camera was progressive scanning type, and had a random trigger function, VGA class resolution, and RGB analog outputs, while a direct lighting system (DL) was adopted as its illumination device.
The grading robot graded 9 fruit varieties of peaches, pears and apples into 5 levels based on extracted features on color, shape, disease and insect injury. From the robotic grading operations, it was observed that there was little difference among grading results by 4 robots. Judgment results of the robots were similar with those of human, because criteria of the robot judgment were set for top and bottom half spheres respectively as human.