Differential filtering algorithm for robot welding seam image enhancement

Abstract

The low resolution of the welding seam recognition image during robot welding will affect the welding accuracy, and the cost of high-resolution cameras is higher. Therefore, in this paper, the geometrical processing of the bilinear interpolation method for the welded seam pictures taken by the low-resolution camera was used to obtain a high-resolution image. This paper proposed a differential filtering algorithm based on Gaussian filtering and bilateral filtering, to eliminate the fine lines, noise and enhance the weld feature value for the obtained high-resolution image, thereby ensuring the welding accuracy. Finally, through the image processing experiment of increasing the 1.25 million pixel weld image to 5 million pixel weld accuracy, the effectiveness of the image enhancement method proposed in this paper was proved.