A Virtual-Reality Occlusion Processing Method based on Dynamic Transformation Background Frame

• 1、Institute of Web Intelligence, Department of Computer Science and Technology, Chongqing University of Posts and Telecommunications, Chongqing 400065, China

Abstract：For getting better sense of reality in Augmented Reality systems, the occlusion relations between virtual objects and real objects must be handled correctly. Since the existing background clipping algorithms are only applied to fixed background and their accuracy of depth recovery for monocular camera are not enough, a virtual-reality occlusion processing method is proposed based on dynamic transformation background frame. This method detects moving objects and gets the contour of foreground objects through transforming the background frame dynamically. Then it traverses the image pixels to determine the occlusion relation between virtual-reality objects, and uses template caching technology to remove the blocked parts. The experimental results demonstrate that the proposed algorithm can be simply implemented, supports moving camera. It has accurate occlusion judgment and less calculation, and is good at dealing rigid and non-rigid object.

Keywords： Augmented Reality virtual-reality occlusion contour extraction template testing