SPATIOTEMPORAL SEGMENTATION OF RANGE IMAGE SEQUENCES INTO PLANAR SURFACES FOR COLLISION AVOIDANCE

This paper presents a spatiotemporal segmentation algorithm for range image sequences of traffic scenes. The technique uses an optical range camera that leads to segmenting the range image sequences into planar surfaces. Vehicle and road recognition can be performed by applying perceptual grouping strategies to merge the detected planar surfaces into objects. Results of the algorithm are shown for two types of range data.