Greyscale image fusion plays a crucial role in computer vision and image processing. This paper proposes a novel algorithm for greyscale image fusion that enhances clarity and information retention of the resulting image. Firstly, the source image is decomposed using discrete fractional wavelet transform (DFRWT) to acquire the low and high frequency components. Secondly, considering the different characteristics of low and high frequencies, the spatial frequency technique is chosen to address the coefficients in the low frequency domain. In the high-frequency part, an improved chaotic genetic algorithm (ICGA) is used to optimize the high-frequency fusion coefficients, which can adaptively adjust the weighting to avoid the blurring and distortion problems that are likely to occur when the DFRWT processes the details, as a consequence, the features of the original image are better retained, making the fused image more natural and the details clearer. Finally, the inverse DFRWT is performed to get the fusion results. Simulation experiments prove the effectiveness of the new algorithm.