An improved linear depth model for single image fog removal

Outdoor images lose color contrast and visibility in poor weather conditions (like fog, mist, haze and rain), which affects computer vision applications extremely. Degree of degradation at a pixel varies with the depth of a scene point from the observer. Therefore, the problem of image restoration under bad weather is expressed as depth estimation of each scene point from degraded image. The proposed work introduces a linear depth model based on color attenuation prior to estimate depth of each scene point from a single image. The proposed work is based on an observation that the difference of saturation from brightness and hue increases with scene depth and preserves structural similarity of degraded image. The proposed work is capable to preserve the existing edges and recover both the scene depth and the degraded edges. Effectiveness and accuracy of the proposed method is measured qualitatively and quantitatively. The experimental result analysis proves that the proposed method outruns in comparison to the live state of art methods.