Blind Haze Separation

摘要: Outdoor imaging is plagued by poor visibility conditions due to atmospheric scattering, particularly in haze. A major problem spatially-varying reduction of contrast stray radiance (airlight), which scattered the haze particles towards camera. Recent computer vision methods have shown that images can be compensated for haze, and even yield a depth map scene. key step such scene recovery subtraction airlight. In particular, this achieved analyzing polarization-filtered images. However, requires parameters These were estimated past studies measuring pixels sky areas. This paper derives an approach blindly recovering parameter needed separating airlight from measurements, thus contrast, with neither user interaction nor existence frame. eases image dehazing, also compensation attenuation. The has proved successful experiments, some are here.