Focusing on infrared and visible image fusion of the same scene,a new image fusion algorithm based on nonsubsampled contourlet transform（NSCT）and improved pulse-coupled neural networks（IPCNN）was proposed.Firstly image was decomposed sparse with various scales and directional features using NSCT,and then a calculation matching selection principle with the varieties of directional bandpass subband coefficients based on IPCNN was developed.Compared with other methods,the results of experiment show that the proposed algorithm can infuse important information of infrared and visible image to fusion image,acquire better results,and improve quality of fusion image.