Abstract: In order to overcome the shortcomings of existing image defogging methods and further improve the effect of image defogging, an adaptive image defogging method based on quadtree decomposition was proposed. According to the standard deviation, the original fog image was decomposed into an iterative quadtree. According to the feature that the sky area was bright and smooth, the median value of the pixel in the area with the largest difference between the mean value and the standard deviation was selected as the atmospheric light. According to the mean and standard deviation of the brightness of the region, the transmission rates of each region were calculated in an adaptive neighborhood manner. According to the estimated atmospheric light and transmission rate, the fog free image was estimated based on the atmospheric scattering model. Finally, the brightness of the fog free image is optimized by the adaptive gamma function. The experimental data show that, the method proposed in this paper has a better visual effect after image defogging compared with the newly proposed defogging methods. The average gradient and information entropy of the defogging image are respectively 6.56% and 1.75% higher than those of the existing methods respectively, while blind/referenceless image spatial quality evaluator is 8.25% lower than the existing methods. This research is helpful to visual detection, recognition and tracking.