A wrist -mounted robotic vision system for recognition and location of three-dimensional objects
Abstract
This thesis deals with the key unresolved issues in robotic application of computer vision: hand-eye calibration to allow visual guidance of a robot, and automatic recognition of CAD-like objects to allow the robot to perceive its environment. The hand-eye calibration is solved by a novel method based on the solution to a homogeneous transform equation of the form AX = XB. The visual recognition issue is formulated as a problem of organizing image edges as loops, finding their viewpoint invariant properties, and matching them to model loops. Each loop is segmented as concatenation of individual straight and elliptical segments for easy matching. Position of general loops are calculated by space resection while that of elliptical loops are computed by a novel elliptical back-projection method.
Degree
Ph.D.
Advisors
Ahmad, Purdue University.
Subject Area
Electrical engineering
Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server.