Abstract: This paper presents a new approach for the effectively tracking of welding seam based on CCD(charge coupled device) sensor.Due to the presence of many spatters,dust and strong arc noises in welding environments,it is very difficult to detect the legible welding seam from the CCD sensor.This paper presented a novel approach,based on the Bubble and M-band wavelet transform,to detect the image edge of welding molten pools from the images captured in metal active-gas welding process.The experimental results show the proposed method is effective in detecting the edges of welding molten pools and identifying the welding seams even when the welding images are presented with much noise.Based on the identification of welding seam,a proportional integral differential control approach is employed to manipulate the welding torch in order to track a desired S-shape welding seam.The control experiments demonstrate that tracking an S-shape welding seam is delivered effectively with good precision based on the proposed welding seam tracking.