안전선영역 감시를 위한 적응적 물체 인식에 관한 연구

초록

In this paper, we propose a method of the adaptive object recognition in safe zone monitoring. The proposed method distinguishes an human and object according to the feature of movement and size of them. In the human case, although the transient behavior can be stopped but movement of human is usually changed over time. Meanwhile, almost there is not change in the object case. Therefor it is to be an object that the movement is low and the size does not change, in opposition, high movement and change of size means that human may be detected. The feature's variation controls the update speed of background image. low variation makes speed up, but high variation reduce the update speed. That prevents continuous detection fault(object detection) and if it detects a person, it can be sustained warning broadcast.