Improved Visibility of Single Hazy Images Captured in Inclement Weather Conditions

Abstract

Haze removal is the process by which horizontal obscuration is eliminated from hazy images captured during inclement weather. Sandstorms present a particularly challenging condition, images captured during sandstorms often exhibit color-shift effects due to inadequate spectrum absorption. In this paper, we present a new type of haze removal approach which uses a combination of hybrid spectrum analysis and dark channel prior in order to repair color shifts and thereby achieve effective restoration of hazy images captured during sandstorms. The restoration results and qualitative evaluation demonstrate that our proposed approach can provide superior restoration results for images captured during sandstorms in comparison with the previous state-of-the-art approach.