Accession Number : ADA299377

Title :   Quarterly Progress Report on Contract N00014-93-1-1235 for February 1994 - April 1994 (Yale University, New Haven, Connecticut),

Corporate Author : YALE UNIV NEW HAVEN CT DEPT OF COMPUTER SCIENCE

Personal Author(s) : McDermott, Drew

PDF Url : ADA299377

Report Date : 1994

Pagination or Media Count : 6

Abstract : In the area of visual tracking, Greg Hager and his students have demonstrated the ability to perform distraction- and occlusion-resistant tracking of closed contours. The novel features of our approach are: a foveal/peripheral approach to image processing that is used to detect approaching contours, and a discrete-event system which is used to predict the possible appearances of a feature if an approaching contour were to cross it. Tests of the system have shown it to be extremely robust to distractions (objects moving behind the tracked target) and robust to occlusion provided sufficient unoccluded area remains to maintain contact with the contour. We are currently working to improve the performance of the system, and to incorporate these ideas into other systems using feature tracking. In the area of visual servoing, we have for the first time demonstrated full six degree-of-freedom visual servoing using calibration insensitive feedback algorithms. Based on this demonstration we are designing a more ambitious visual servoing system that will support fast and easy reconfiguration for a variety of manipulation tasks.

Descriptors :   *STUDENTS, *RESEARCH MANAGEMENT, *VISUAL SURVEILLANCE, COMPUTER PROGRAMS, ALGORITHMS, IMAGE PROCESSING, TRACKING, TARGETS, SENSITIVITY, FEEDBACK, CALIBRATION, CONNECTICUT, FOVEA, CONTOURS.

Subject Categories : Anatomy and Physiology
      Personnel Management and Labor Relations

Distribution Statement : APPROVED FOR PUBLIC RELEASE