Weld seam detection and feature extraction based on laser vision

This paper presents an algorithm for real time weld seam detection and feature extraction of butt weld, which is based on laser vision sensor installed on the end hand of welding robot. The algorithm preprocesses the weld seam images captured by laser vision sensor as first step and then detects laser stripe edges thus obtains the center of laser stripe. Then a two-step method is used to extract the initial value of center point of weld seam. Finally the weld region is separated from the laser stripe center line using iterative search method, then two edge points and accurate center point of weld region together with the width of weld seam can be calculated. Experiments have shown that the algorithm proposed in this paper has good stability and accuracy, and it can meet the real-time requirements.