Abstract: The image processing and pattern recognition was first used to obtain information of the TIG welding process of aluminum alloy.The image of welding pool of aluminum alloy was pretreated with a series of methods using weighted median filter,statistical gray enhancement,threshold by expectation and projection according to the strong noise in the image.A neural network method was used to process image of welding pool and the result of detecting the edge of binary image with BP neural network was excellent.The symmetry of welding pool of aluminum alloy was studied when the welding current is great.The whole welding pool edge of the image is obtained in the single side image and the accurate measuring method of the welding pool geometry parameter is provided.