To overcome the weakness that traditional fused image cannot express the image contrast and details well, an image fusion algorithm is proposed based on non-subsampled shearlet-contrast transform (NSSCT). The correlation and diversity between different high coefficients are analyzed by non-subsampled shearlet transform (NSST), and the NSSCT is built with same orientation high coefficients, which reserves contrast and details image information. Based on image characters of lower frequency, low coefficients are fused by enhancing the salient targets of image. Fused image with higher contract and enhanced details is obtained by inverse NSSCT. Compared with several popular algorithms, such as wavelet transform, NSST and saliency map, the experimental results show that the proposed algorithm is obviously of superiority in reserving image contrast and details.