Research in Knowledge-Based Vision Techniques for the Autonomous Land Vehicle Program.

摘要

The authors' basic approach to detecting and tracking motion is to extract and match features, such as lines and regions, from a sequence and to generate motion estimates from these. They present one report on spatio-temporal analysis for tracking edges through very closely spaced sequences. They also present a report on matching edge-based contours using edges from multiple scales with low resolution guiding high resolution matches. They also present an analysis of estimating 3-D motion and structure of moving objects with uniform acceleration. Keywords: Motion analysis; Target detection and description; Knowledge-based vision.