Abstract:

The processing of welding images is unavoidably accompanied with noises due to the dust , strong arc light and other interference factors in the welding process. As the traditional methods of image processing are sensitive to the noise , it is difficult to accurately.process the image. In order to solve this problem , this paper presents a novel composite image processing method based on the fractal and mathematical morphology theories. In this methpd , first , the gray-scale image of the molten pool is obtained according to the fractal-based discrete Brown random field theory. Next , the connected domain is detected based on the binary morphology theory and is then used to remove the noise and obtain the target image. Moreover , the successive edges of the image are obtained by means of the Open-Close operation on the basis of the mathematical morphology theory. Experimental results show that the proposed method is of high speed and accuracy in the recognition of molten pool information including the edge , the height , the width and the area.

Key words: TIG welding, molten pool, image, fractal dimension, mathematical morphology, edge detection