Implementation of haze removal algorithm to enhance low light images

Abstract

The image is captured in foggy atmospheric conditions, resulting in hazy, visually degraded visibility; it obscures image quality. Instead of producing clear images, pixel-based metrics are not guaranteed. This updated image is used as input in computer vision for low-level tasks like segmentation. To improve this, it introduces a new approach to de-hazing an image, the end-to-end approach, to keep the visual quality of the generated images. So, it takes one step further to explore the possibility of using the network to perform a semantic segmentation method with U-Net. U-Net will be built and used in this model to improve the quality of the output even more.