MULTILEVEL- AND NEURAL-NETWORK-BASED STEREO-MATCHING METHOD FOR REAL- TIME OBSTACLE DETECTION USING LINEAR CAMERAS

This article details an approach to solving the problem of real-time obstacle detection using linear stereo vision. The author presents a multilevel neural method for matching edges extracted from stereo linear images. The method described assigns priorities to edge matching and processes them at different levels to save computational time, which is notoriously long in common proactive real-time detection. The authors process selects significant edges to work on computationally and uses their position to assign importance to other nearby edges. Optimization is the key component here, with the author designing a system that pays full attention the most important objects, while discarding excess information about objects of lesser importance.

Language

  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 00989342
  • Record Type: Publication
  • Source Agency: UC Berkeley Transportation Library
  • Files: BTRIS, TRIS
  • Created Date: May 3 2005 12:00AM