Improved Defogging Algorithm Based on Fusion of Dark-Bright Channel and Sky Segmentation

In adverse weather conditions with low visibility such as smog and rain, the large amounts of water vapor and dust particles suspended in the air cause scattering, refraction, and absorption of visible light. This results in images captured by imaging devices having low contrast and color distortion, severely affecting the normal operation of outdoor visual monitoring systems. This paper proposes a rapid image dehazing method based on dark and bright channel priors and adaptive thresholds. First, the dark channel, bright channel, and grays cale images of the hazy image are calculated. The threshold and atmospheric light value are estimated through the bright channel image, and the dark channel image is processed using an improved Sobel edge detection algorithm to create a mask for refining the transmission map. Next, compensation is performed for the sky region to address the block effect problem present in the sky area. Finally, the clear, haze-free image is restored using the transmission map and atmospheric light value. This method effectively improves image contrast and clarity, reduces artifacts and noise problems present in traditional dehazing algorithms, significantly enhances dehazing effects, and meets the application needs of autonomous driving, intelligent transportation, and photogrammetry. Experimental results show that this method outperforms traditional algorithms in terms of dehazing effect, processing speed, and image quality, demonstrating significant practical value and broad application prospects.