기계시각을 이용한 대립종자의 씨눈위치 판정알고리즘 개발

This study was conducted to develop an algorithm for the embryo location in seed by using machine vision. The topic of this research is to detect the embryo location in seed regardless of seed supply direction. In order to detect the embryo location in Cham Bak, Tuktojwa and Hukjong, the effect of seed posture in the supply line was investigated. When the seed posture angle of Chambak from horizontal direction was 30°, the detection accuracy for embryo location was 77.8%, while detection accuracy was 100% for the 0° or 15°. When seed posture angle of Tuktojwa was 30° from the horizontal direction, the detection accuracy was 89.5% and it was 100% for the 0° and 15°. Embryo location detection accuracy for the Hukjong was 94.4% when the seed posture angle from the horizontal direction is 30°, and it was 100% for the 0° and 15°. When seeds are fed into the posturing and seeding line, the seed postures within 30° with mechanical means, and at most 15° seed stand posture, the developed algorithm can detect the embryo position in the seed. So, this embryo detection system is very useful tool in the posturing and seeding line.