A New Approach for Dehazing and Enhancement of Infrared Images

This paper proposes the implementation of an algorithm that dehazes and enhances outdoor infrared (IR) images. Dehazing is carried out through dark channel prior (DCP) and enhancement is achieved by multi-scale retinex (MSR) technique. The atmospheric light present in the haze concentrated image is approximated and a transmission map is estimated which separates out the concentration of haze. Enhancement of dehazed images is done by augmenting the image contrast directly instead of unraveling the physical model of image degradation. The reflectance component of image is acquired by approximating the atmospheric brightness of scene. Resulting images are then assimilated in optical attenuation model to recover the final output image. The proposed method is implemented using MATLAB and is validated by assessing the images according to their average gradient, information entropy, and root mean square error. Experimental results show that the proposed method can remove the haze concentration in images successfully and enhance them significantly.