テンプレートマッチングに色差エッジ画像を用いた背景差分法による警備ロボットの空間変化検知

抄録

In this paper, we focus on robotic patrolling systems. Patrolling robots are required to detect changes in environments due to suspicious and missing objects. We have thus far proposed a background subtraction method using two color difference-based edge (CDE) images. However, it is difficult for autonomous mobile robots to acquire the images at the same positions because of the localization error and collision avoidance with obstacles. For the input images in which the backgrounds are different from the standard one, the detection accuracy is decreased. For this problem, we propose to use template matching in the background subtraction method. In the experiments, the patrolling robot successfully detects not only suspicious objects, but also missing objects. Furthermore, the detection accuracy is increased and processing time is reduced compared to a conventional background subtraction method using color edge images. From the results, finally, the effectiveness of the proposed method for the spatial change detection by the patrolling robot is shown.