Occlusion Culling for Image-Based Rendering withWarping

Abstract

The basic idea of the proposed algorithm is to reduce the number of depth pixels in Layer Depth Images (LDI) by culling the occluded ones before warping. The method combines the octree spatial subdivision concept with LDI concept, using the implicit geometry stored in depth pixels. The algorithm uses an octree to group depth pixels in a hierarchical way. The proposed algorithm is very effective when used in highly occluded scenes, and the density of the depth pixels should be high enough to get a high performance from the algorithm