Aimed at the problem of detection of pool edge for aluminum alloy MIG welding, a new method is adopted. The method is used to detect the pool edge of aluminum alloy MIG welding by applying the pulse coupled neural networks algorithm. The pulse coupled neural networks algorithm is introduced in this paper. The detection of pool edge for MIG welding pool is carried out with this algorithm by MATLAB. Then, the comparison between the result of PCNN algorithm and Canny detector is completed. The result indicates that the pulse coupled neural networks algorithm can be used to detect the process image of welding pool. And the image of pool edge by using the pulse coupled neural networks algorithm is clear and consistent. Besides, this new method can effectively overcome the influence of noise from welding pool image. The time of single frame processing is 54 ms.