Aiming at the defects of the Retinex algorithm, such as poor edge preservation, easy to produce halo and over-enhancement, and the phenomenon that bilateral filtering is easy to cause the loss of detail of the image, an image enhancement algorithm based on fusion idea under Retinex theory is proposed. The algorithm firstly extracts the luminance Y component in the YCbCr color space, and performs MSR enhancement on the large, medium and small scales of the luminance component to obtain a brightness image with better information preservation. At the same time, the single-scale Retinex enhancement based on bilateral filtering is performed in RGB color space to obtain the enhanced image with better edge information. Then, the enhanced two images are weighted and fused. Finally, the fused image is processed by color restoration to obtain the final enhanced image. Compared with the images processed by this algorithm, the classical SSR, MSR and MSRCR algorithms, the experimental results show that the images processed by this algorithm are superior to other algorithms in details, color and edge preservation, and avoid the occurrence of halo and over-enhancement.