Due to uneven lighting at night and the complex environment, the current nighttime dehazing algorithm has problems such as light source diffusion, color distortion after dehazing, and bad image quality. According to the characteristics of haze images at night, this paper proposes a nighttime dehazing algorithm based on mixed filter light estimation and transmission optimization. Aiming at the inaccurate estimation of the ambient light at night, it is proposed to first perform side window filtering on the brightness image to ensure the illumination direction, and then use the guided filtering to refine the three channels as the local ambient light estimation; to solve the problem of light source diffusion caused by the application of dehazing at night, High-light area compensation is used to improve the transmission of the light source area. At the same time, for the uneven color and loss of details after dehazing, it is proposed to use guided filtering to correct the coarse transmission and then regularize the solution. Finally, the dehaze image is solved by the atmospheric scattering model. Experimental results show that the algorithm achieves good dehazing effects at night: in subjective evaluation, the algorithm can dehaze on the basis of preserving the original color of the image, improving image details and effectively suppressing light source diffusion; In terms of objective evaluation, all evaluation indicators of the algorithm have been greatly improved.