Qualitative Decomposition of Range Images into Convex Parts / Objects

Previous works on range image segmentation concentrate on surface patches that can be well represented by certain mathematical functions. In this paper we consider the more qualitative segmentation problem of decomposing a range image into convex parts/objects. An edge-based approach is proposed which uses an adaptive contour closure algorithm in conjunction with a global convexity test. Experimental results on two range image sets are reported to demonstrate the performance of our convex decomposition technique.

• Type: Conference
• Year: 2000
• Where: MVA
• Authors: Xiaoyi Jiang
• Keywords: MVA 2000, MVA 2007, Qualitative Segmentation Problem, Range Image, Range Image Segmentation