In a high-power fiber laser welding process, the thermal radiation of a weld pool contains plenty of information for welding quality, in which the pool width can reflect the welding stability. Thus, extracting the welding pool width of high-power fiber laser welding based on infrared thermal imaging is an important method for monitoring the weld seam quality. In this paper, we studied the 304 stainless steel welded by a 10kW high-power fiber laser continuously. A near-infrared high-speed sensing camera was used to capture the weld pool images. Image algorithms such as median filtering, gray scale stretching, cutting, dynamic threshold mathematical morphology were applied to extract the weld pool image edge, analyze and detect the weld pool width. Welding experimental results showed that the proposed methods could extract the weld pool width, which could reflect the stability status of high-power fiber laser welding process accurately.
