An fusion algorithm for multi-spectral and high-resolution images based on multi-level threshold segmentation and the nonsubsampled contourlet transform (NSCT) is proposed. Multi-level threshold segmentation is done for the multi-spectral image, and notions about ratio of region mean (RRM) are used to divide the multi-spectral image into the areas which need to be spatially enhanced and need to preserve spectral characteristics. Then the NSCT is performed on the high-resolution image and the intensity component of the multi-spectral image at different scales and directions. The low-frequency coefficients are fused with the windows-based fusion rules and operators, and the high-frequency coefficients are fused with the region-based rules according to the RRM. Finally the fused coefficients are reconstructed to obtain the fused intensity component of the multi-spectral image, and the inverse IHS transform is performed to get the fused multi-spectral image. Experimental results show that the algorithm proposed performs significantly better than the IHS transform, the pixel-based à trous wavelet transform and the pixel-based NSCT.