MACHINE VISION TO INSPECT TOMATO SEEDLINGS FOR GRAFTING ROBOT

Abstract

Nearly 60% tomato seedlings are grafted in Japan. However, almost all of them are being done manually, which is labor-intensive, tedious and expensive. In order to develop a fully-automatic grafting robot for tomato seedlings, the machine vision system for sorting grafting seedling and inspecting grafted seedlings was studied. In this paper, the images acquired with selected devices and different lighting (Front lighting, Backlight, and Front lighting & Backlight) were processed based on the theory of machine vision. The experimental result shows that a blue backlight without a filter device is the best to inspect the seedlings. The images of inspecting seedlings were captured using the selected machine vision system. The algorithm developed using the machine vision system was found functional in inspecting the grafted seedlings; it can even detect a very small gap (≤0.2 mm) between the scion and rootstock after they have been grafted.