Weld seam feature point extraction using laser and vision sensor

Vision sensors are used in welding seam applications for image capture and post-processing of image data. When utilized with welding robots, it is essential for establishing control quality of seam tracking that is up to par. The vision sensor works with a laser to provide the advantage of extracting the location of the weld seam. However, in terms of visual perspective, the challenges may come from the possible environment illumination, imperfections on the workpiece surface such as stain and scratches, error from the hardware setting, or laser reflection which produces a lot of image noises. The noise generated by the laser beam on the surface of the workpiece due to the specifications of the laser itself can cause the centerline of the laser to become uneven, usually making it difficult to extract the feature point of the weld seam. In this work, a method to improve the quality of the centerline by constructing a more stable centerline has been presented to obtain the pixel location of the welding seam point in the x–y coordinate by considering the effect of laser reflection on the workpiece. Three main phases have been employed to realize the objectives: laser extraction, broken-line fitting, and pixel location determination. To locate a pixel coordinate, a technique of the qualitative description based on the V-groove of butt joint had been developed. Two metals of butt joint, each of which has two different types of lines with several laser positions on the workpiece are tested in order to validate the accuracy of the proposed method. The approach has minimal error outcomes when compared to the actual value and has no noticeable impact on the system’s effectiveness.