Non-linear Root-signal Fusion based Single Hazy Image Enhancement

Abstract

This paper presents a new technique for enhancing hazy images taken in indoor, outdoor, or underwater imaging environments. In the proposed technique, several root-signals having different bandwidths are derived from the given hazy image and its complement image. These root-signal images are enhanced through non-linear enhancement techniques using adaptive histogram equalization. The resulting images are fused to give the final dehazed image. The proposed technique is simulated using MATLAB software. The proposed methodology results are presented, and it is observed that the subjective and objective quality assessments of the dehazed images are comparable to or better than some of the existing state-of-the-art methods.