Layer Based Approach of Corner Detection and Classification in Quantized Range Images

This paper deals with corner detection and classification in low depth resolution range images. Having significantly lower range resolution than spatial introduces challenges in spatial information understanding. The method consists of processing the layers individually. Each layer is treated as binary images and morphologic operations are applied to extract iso-depth lines. Based on these lines we detect and classify corners using a set of conditions. We present results of the algorithm based on synthetic images generated from different viewpoints of a scene.