Polarimetric dehazing utilizing spatial frequency segregation of images.

A procedure for the detection and removal of haze from dense hazy images has been proposed. It involves the analysis on the content of low-spatial-frequency information of a scene. The image contaminated by haze is decomposed into different spatial frequency layers by the wavelet transform, by which the hazy parts of the image are focused on the low-frequency components. A dehazing method combining both the airlight and direct transmission is employed to specially dehaze the low-frequency parts. The high-frequency parts are processed by a transfer function to enhance the clarity of the hazy image. Finally, a dehazed image with high clarity is obtained by image construction which employs the low- and high-frequency components. Experiments and analyses demonstrate the good performance of the scheme in terms of improving the contrast and clarity of hazy images. Particularly, it works well in improving the visual range of images captured in hazy weather conditions.