Nighttime Haze Removal using Multi-Scale Exposure Fusion

Abstract

Outdoor images sometimes suffer from bad weather conditions which decreases the quality of images. In image processing, the technique used to recover a good quality image from such a damaged image is known as Image Dehazing. In this paper, a new approach for image dehazing is proposed based on fusion technique. The proposed model enhances the image quality using a series of operations like white balancing, contrast stretching and multi-scale Laplacian blending scheme. The input image is given to the standard Shades of Gray algorithm in order to perform white balancing process and simultaneously contrast stretching is performed on the same input image. Both the white balanced and contrast stretched images are then given to their respective exposure fusion algorithm wherein fusion procedure applied to white balanced image is slightly different than that of contrast stretched image. Thereafter, Brightness enhancement of the output image is done by estimating the brightness using original image to get enhanced haze free image. Obtained results indicate that this multi-scale fusion-based approach not only gives better results but also maintain color of the original image.