Abstract: In order to address the issues of inaccurate transmission estimation, poor color restoration in sky regions, and slow operation speed in current defogging algorithms, this paper proposed a real-time video defogging algorithm based on multi-scale guided filtering. First, the automatic white balance algorithm was used to correct the color of the foggy image. The confidence factor of the color corrected image and the dark primary color were used as the guiding filter. Then, the pyramid sampling technology was used to obtain the reduced image. Then, the quadtree method was used to estimate the transmittance and atmospheric light intensity, and the iterative up sampling and guiding image filtering were continuously used to effectively prevent the loss of information and finally achieve the best transmission effect. In addition, we extended the single image defogging algorithm to real-time video defogging, reducing flickering artifacts in defogging videos by ensuring consistent transmission values over time. The experimental results show that the algorithm has a fast running speed, significant defogging effect, and less flickering artifacts in video defogging. Compared with the shielded Poisson equation defogging algorithm, this algorithm has faster time 81. 32% and higher average gradient 67. 08%, which meets the requirements of fast and stable real-time video processing.