Considering the features of infrared image and low-light-level image, an image fusion and colorization method based on sparse representation and colour transfer is proposed in this paper. Firstly, the blocked sparse representation method based on K_SVD is used to extract image features and then the grayscale fused image with the max fusion rule is gained. Secondly, the colour is transferred by matching the statistics of target image and the reference image in the YUV colour space. Finally, the final color fusion image is obtained by replacing the luminance component with the grayscale fused image during the color transfer process. The experimental results show that the proposed algorithm can integrate the features of infrared image and low-light-level image well. And this method can obtain the perfect contrast and detail information, and the colour of the fusion image is comfortable for human eyes.