Study of weld pool monitoring system based on spatial filtering

The observable information of the welding process mainly includes the weld pool, weld seam, and arc length, which can be used to predict the penetration depth, perform weld tracking, and provide early warning of weld defects. Welding monitoring technology based on machine vision has been widely researched to improve the quality and efficiency of welding. This paper proposes a novel weld pool image acquisition system based on spatial filtering and Abbe’s theory of image formation. A high-pass spatial filter was designed to eliminate the low frequency background noise so that the contour of the weld pool, weld seam, and arc shape were acquired. Compared with the weld pool monitoring system based on an optical filter, this method more effectively solved the problem of the large radiation gradient in the weld pool area, and the weld pool could be more clearly observed. It provided a new method to image the weld pool image system with the advantages of low cost and ease of implementation. • Weld pool imaging's process is analyzed based on Abbe's theory of image formation. • A high-pass spatial filter was used to change the spatial spectrum of the weld pool. • The spatial filter effectively solves the large radiation gradient in the weld pool area.