主催: 一般社団法人 日本機械学会
会議名: ロボティクス・メカトロニクス 講演会2018
開催日: 2018/06/02 - 2018/06/05
While many convenient home appliances have spread in Japan, dishwashers are not widely used. One of the reasons is the difficulty of loading dishes to a dishwasher. Thus the authors' group proposed a system to support placement of dishes in dishwashers. The system counts the number of dishes from a table image after a meal, and optimizes the layout of the dishes in a dishwasher. In the dish-counting subsystem, we adopt template matching using SURF local features. This paper addressed improving the counting accuracy of the subsystem. First, by subsampling correspondences of the SURF points using color invariants, the number of erroneous correspondences is reduced. Secondly, we can get the area of the dish on which SURF points exist using their scales and orientations. In conclusion, the accuracy was improved than our previous method.