Line Correspondences from Cooperating Spatial and Temporal Grouping Processes for a Sequence of Images

摘要

This paper addresses the problem of matching line segments in two images from a sequence of images. We propose a new algorithm that employs relaxation labeling to integrate the perceptual grouping and feature matching processes. We consider feature matching between two views as a “temporal grouping” process, in addition to the traditional spatial groups established by perceptual grouping in a single image. In the relaxation procedure, we compare a line in the first image with multiple matching candidates in the second image. Correspondence ambiguities are resolved by iteratively propagating constraint information from the neighbors of a line, which are defined by the line's perceptual groups. The applications of our work include moving object detection, image registration, mobile robot navigation, and 3D structure from motion. We discuss the new algorithms and present performance analysis and experimental results with real image sequences.