깊이 영상 분류 기법을 이용한 테이블탑 상호작용 시스템

초록

In this paper, we propose an efficient tabletophuman-object interaction system by categorizingdepth information. This system uses three layers toapply proper algorithms respectively. Backgroundlayer is the trained background environment modelwhich is initialized at the beginning. The staticlayer is used to store steady objects which isregistered for interaction subject. And the dynamiclayer handles moving objects which in this case ishand. In each layer, a proper algorithm can beapplied separately without interference. The systemwill show how effective this approach works.