Single Image Haze Removal Algorithm with Upgraded Transmission Map Model

Abstract In a single image, haze/fog removal is an exigent problem due to its ill-posed character. In general, the efficiency of haze removal algorithm depends on the accurate estimation of the transmission map and the perfection of this map relies on the estimation of scene depth and scattering coefficient. Existing methods focus only on the improvement of scene depth model and keep scattering coefficient as a constant. However, scattering coefficient equally contributes in the calculation of transmission map because haze density varies with the scattering medium. Therefore, the variable scattering based transmission map is required for perfect haze removal algorithm. In this paper, the upgraded transmission map is proposed which is based on the variation of scattering information in foggy image. Apart from this, in the proposed model, linearity is achieved by the ratio of difference to sum of intensity and saturation. Besides, proposed method also take care the well known problems of edge preservation and colour fidelity. Experimental results manifest that the proposed model provides better efficiency as compared to the existing haze removal algorithms in terms of qualitative and quantitative analysis.