抄録
We studied high speed detection of shape characteristics from range image for virtual plant factory. The range image was measured using a laser range finder. In this application, we used a laser sensor that was more wide ranged than that used in previous study, because we measured larger fruits. We made a 3-dimensional model of the fruits that was constructed by a plane represented by polygons.
We made an algorithm for real time processing to find the characteristic points of the polygons. This algorithm had a parameter to cope with different shapes of plants, and it was decided based on the type of plant. In this study, we found the reasonable parameters to construct the polyhedron. The surface of the polyhedron was just like that of the fruits. Processing time of the polyhedron was usually about 2-5 sec. and it was less than 10 sec. even on the fruits with complex shapes. This algorithm was so fast that we could use it for real time operations.