Hazy image and video enhancement using multi-scale guided filtering

Abstract

In this article, we propose an algorithm for enhancing the quality of images and videos captured under foggy weather conditions. First, the input foggy image is downscaled to construct an image pyramid. After that, we estimate the transmission map and atmospheric light at the coarsest layer of the pyramid. Next, the transmission map at the finest layer is obtained by sequentially applying the guided image filter across the pyramid’s image layers. Based on the collected information, we refine the transmission map for higher-resolution layers by applying a guided image filter progressively through the pyramid. This hierarchical approach ensures efficient computation while preserving essential image features. By incorporating real-time processing techniques, the proposed method significantly reduces flickering artefacts, which often degrade the visual quality of defogged videos. Experimental evaluations demonstrate that it can be implemented in real-time applications while achieving performance comparable to or superior to state-of-the-art defogging methods. This makes the approach highly practical for scenarios requiring visibility enhancement.