An Image Restoration Method for Outdoor and Its Application to Under Water Using Improved Transmission Map and Airlight Estimation

Abstract

Dehazing is an important image restoration technique to remove the presence of Haze from a hazy image. Recent dehazing algorithm is not sufficient to remove Haze from the given outdoor or underwater hazy images. Therefore, an efficient dehazing algorithm is needed for the removal of Haze. Initially, multiple image dehazing methods are used to remove Haze and these dehazing methods have many drawbacks such as, multiple image methods cannot be applied to dynamic scenes and cannot provide results instantly. In order to overcome drawbacks of multiple image dehazing methods, Single image dehazing methods are introduced which are based on some important observations or priors. One such single image dehazing technique is dark channel prior. The thickness of Haze and airlight is estimated using dark channel prior. Guided Filter technique is used to refine the transmission map. But the estimated Haze thickness is inaccurate because of the usage of minimum operator in dark channel prior method. To improve the estimation of Haze thickness, the edge collapse based repair is used after dark channel prior and guided filter technique. This paper presents the time-efficient dehazing of outdoor images with patch size of 25 × 25 and airlight of 3% and this principle is applied to remove Haze in underwater images. The experimental result shows a better result for both outdoor and underwater images.