Development for Vegetable soybean (edamame) sorting machine to use image processing

Abstract. Green vegetable soybean (Glycine max), also known as edamame in Japan, is an increasingly popular nutritious food in eastern Asia and the United States. The main producing prefectures in 2013 were Niigata (1,580 ha), Yamagata (1,460 ha), Gunma (1,150 ha), Akita (1,080 ha), Chiba (854 ha), and Hokkaido (752 ha). These prefectures accounted for 55% of all the planted area in Japan. Edamame cultivation is done using advanced power farming systems in Japan. Nevertheless, sorting, the final processing of edamame, is time-consuming: about 12 kg/hr or 48 hr/10a are necessary when using manual labor. Because even edamame is harvested at the optimal time, 36–44% of pods are classified outside the grade. Among them, 6–15% show mechanical damage. Therefore, sorting constitutes the main hurdle associated with edamame cultivation. It impedes the increase of cultivation area and limits profitability. Therefore, we developed a sorting machine to improve edamame processing efficiency. It comprises image processing units, belt conveyors, and sorting systems with an air blaster. Work rates of developed edamame sorting machine (type-14) were 73–85 kg/hr, which were 8 times higher than those associated with manual sorting. Image-processing units detected seed maturity and edamame damage simultaneously. Sorting results which use type-14 showed average sorting accuracy indexes (η=η_a+η_b-1) of 0.20. This sorting accuracy was 0.10 point better than type-13. The edamame sorting machine accuracy reached 0.55 for good pods (η_a) and 0.63 for bad pods (η_b), each of which was inferior to manual sorting.