Abstract: Weld pool geometry is a crucial factor in determining welding quality and its feedback control is a fundamental requirement of automated welding.Therefore,a vision sensing system for taking and processing the image of metal inert gas(MIG) welding pool of aluminum alloy is developed.Clear images of the MIG welding pool are obtained by using the technology of narrow band filter,which is used to overcome the influence of strong arc light.Meanwhile,the aluminum alloy characteristic of the MIG welding pool is analyzed.In order to overcome the disadvantages of traditional welding pool image processing method,the morphological operators is used to process the aluminum alloy MIG welding pool.The method can effectively wipe off noise and 'cathode pulverization' area in image.Experiments show that the method is effective and feasible,and it can provide reliable information for the real time control of aluminum alloy MIG welding process.