Studies on Sorting of Green Tanba Soybean (Part 1)

Discrimination of the number and kinds of beans in the pod and the distorted soybean using shape analysis

Abstract

The purpose of this study is to develop a soybean sorting system capable of detecting distorted soybean and of detecting the number and kinds of the beans in the pod. Top view of the soybeans was used for image processing. The method of using the centroid radius with peak adjustment was contrived for detecting the long axis, which comprises stem and tip. The centroid displacement was used for the parameters to detect the distorted soybeans. The relationship between the long and short axes of the soybean is used to determine the number of beans. The beans were correctly identified at 98.8%. The distorted soybeans were correctly identified at 100%.