Dehazing Method Through Polarimetric Imaging And Multi-Scale Analysis

Images captured in haze weather conditions are often plagued by poor visibility, short detection distance and decrease of information due to atmospheric scattering. It is always a hot and promising topic in how to efficiently remove the influence of haze and get useful information which users are interested in. Aiming at recovering a haze-free image, we studied the atmospheric scattering model, explored the polarization effect in imaging process, and then analyzed the spectrum characteristics of haze images. Therefore, a dehazing method using polarimetric imaging and multi-scale analysis is presented. The major researches of this paper are as follows.The attenuation model and the airlight model of imaging in atmosphere have been described. Moreover, Haze-images degradation model has been established by simplifying the atmospheric scattering model.The polarization properties of light and ways to detect it have been described. Principle of traditional polarized-based dehazing method and several commonly used algorithms have been analyzed.A conclusion that haze mainly distribute in low-frequency while they weakens high-frequency component of images has presented after a discussion of both spatial and frequency domain features of haze images. Therefore, a novel polarization dehazing method which processes the low and high spatial frequency components separately is proposed. We process low spatial frequency components with a polarization-based dehazing algorithm, while the high-frequency components are manipulated by a nonlinear transform. Finally, the two types of components are combined for reconstruction of high-clarity dehazed images.A new polarization hazy imaging model is presented, which considers the joint polarization effects of the airlight and the object radiance in the imaging process. It fixes the color distortion problems of planar scenes when using the traditional polarized-based dehaze algorithm.Experimental results verify that the dehazing method proposed in this study can strongly promote image visibility and increase detection distance through haze. Moreover, this algorithm is also proved to be useful for preserving image colors. The algorithm can be widely used in outdoor machine vision fields such as target detection and recognition, aerial remote sensing, etc., and can be a guide for underwater image analysis, turbid media imaging. Future research will focus on new approach of blindly recovering with no sky region and real-time polarimetric detection system in order to adapt to future applications of video surveillance and navigation.