An Optimized Image Fusion Method for Fume Removal in Automated Welding Robots Field

Welding is a commonplace process in many industrial sectors. Once the welding environment can be harmful to human’s health, researchers are presenting solutions to automate this process. Many of these solutions use robots guided by computer vision (using cameras). Fume produced by the arcwelding process usually adheres to the camera lens and affects the robots’ perception. Once fume adheres to the lenses or protective glass, none of them can be changed until the whole welding process is completed. The impossibility of changing protective glass makes restoring the acquired image even more important. In this paper, we propose an optimized method to minimize the interference of the fume present in the acquisition system. The solution presented is based on a fusion of different image processing methods. The results show the method is able to improve the groove detection allowing better welding.