Research on fast and accurate occlusion detection technology of augmented reality system

Augmented reality is a rising Human-Machine interaction technology developed from virtual reality technology. It merges virtual information into real scene seamlessly, rather than replace reality completely. The occlusion relation between virtual object and real objects is of great importance to the reliability of rendered merging scene. Currently, most occlusion processing methods are based on deepness calculation, whose execution cost is great. In this paper, we introduce a new occlusion detection algorithm by judging the geometry relation between projection of virtual points and projection of real points according to the theory of stereoscopic geometry instead of by deepness calculation. An optimizing algorithm on triangle mesh occlusion detection is presented for instance when occlusion status detecting is failed because of the matching failure of trianglepsilas three vertices. And also precise epipolar constraint solving method is included to construct a fast, accurate and robust occlusion detection algorithm in augmented reality system.