Masters Theses

Author

Dara T. Khani

Date of Award

6-1988

Degree Type

Thesis

Degree Name

Master of Science

Major

Electrical Engineering

Major Professor

Dragana Brzakovic

Committee Members

A.O. Bishop, R.E. Bodenheimer

Abstract

A vision system for processing multi-pass arc welding images to aid in the automation of the welding process is presented. Information pertaining to the weld groove edges and the weld pool are extracted from a sequence of images acquired through coaxial viewing. The vision system first locates small regions within each frame of the sequence which contain the weld groove edges. These regions are then processed with an edge detector and, based on the results, the location of the weld groove edges are determined. Two processing methods have been investigated to extract the weld pool. The first method locates the region m each frame where the weld pool is present. Next, the edges contained in this region are extracted with an edge detector. The edge information is then processed by a rule-based system which makes use of the weld pool characteristics to determine the weld pool edges. The second method extracts the weld pool by modeling the weld pool as an ellipse and utilizes a transform to map the weld pool into a vertical line. This method simplifies the problem of detecting the boundary of the weld pool.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."

Share

COinS