An image fusion method based on the potential low-rank representation and cross bilateral filtering was proposed. Firstly， the latent low-rank representation method decomposed the source image into low-rank partial image and salient part image， calculated the weighted average of the low-rank part， and summed the salient part. Then， the calculated low-rank part and salient part were used as the input of the cross-bilateral filter， and the corresponding weight was obtained by measuring the detailed intensity. The input image was fused by the weighted average method of cross bilateral filter and smoothed by Gaussian filter to produce the first part of the image. Secondly， the source image used the CBF to obtain the fusion image， and then passed the Gaussian filtering processing to obtain the second part image. The two parts were finally weighted and summed to obtain the final fusion image. After experimental comparison， the proposed method could effectively reduce the rate of gray change of the fusion images and improve the brightness and contrast of the fusion images， which could assure the fused image better observability.