Improving Image Dehazing Performance of Outdoor Scenes Using Contrast Enhancement Techniques

Image dehazing, a crucial computer vision task, aims to improve hazy images by removing the degradation brought on by atmospheric haze. However, the complexity of haze frequently limits the performance of traditional dehazing models, making it challenging to precisely figure out a transmission map approximation. The study involves a novel approach to augment the performance of image dehazing by combining various contrast enhancement techniques with a state-of-the-art dehazing model FFA-Net. The effectiveness of contrast enhancement techniques such as CLAHE, AGCWD, BBHE, DSIHE is investigated on the quality of the dehazed images. Then these techniques were combined with the popular FFA-Net model to form a new hybrid dehazing model. The experimental findings demonstrate that, with respect to no-reference picture quality metrics NIQE, PIQE, BRISQUE, and Average Entropy, the proposed hybrid model of CLAHE+FFA-Net slightly improves the current dehazing approach by slightly improving the Average Entropy and BRISQUE values, while also maintaining satisfactory NIQE and PIQE values.