Image descattering and absorption compensation in underwater polarimetric imaging

Abstract Underwater image contrast degrades and color distorts due to wavelength attenuation caused by scattering and absorption, and the attenuation rate of different wavelengths is significant different in variety of underwater environment, which leads to the satisfying underwater image restoration very difficult. We propose a novel underwater polarimetric image restoration method that consists of descattering and absorption compensation, which particularly takes into account wavelength- and scenedepth- dependent absorption existed in underwater signal light that is ignored by current polarimetric descattering methods. In the descattering step, an automatic attenuation difference map based background region selection strategy is presented, which can estimate background light without considering the no-object region exists or not. In the absorption compensation step, a novel absorption compensation constraint strategy is introduced into the color restoration algorithm for the detected signal light. The strategy relies on well-preserved color component and an absorption prior constraint. Extensive experiments on real-world underwater images demonstrate that the proposed method is more effective than several previous underwater image restoration or enhancement works.