Visibility enhancement of hazy images based on a universal polarimetric imaging method

摘要

Enhancing the visibility of images taken in the hazy weather is important in many applications. Among many dehazing methods, those based on polarimetric imaging techniques have several advantages, such as ease of keeping detailed information, low cost, and high efficiency. In this paper, we propose a novel and robust dehazing algorithm based on polarimetric imaging. By introducing the orientation-angle information from the Stokes matrix, all the parameters used in dehazing performance can be effectively, precisely and automatically estimated, and no additional human-computer interaction is needed. Besides, this method can also be used in handling hazy images without sky region. Experimental results show that such a method can greatly enhance the visibility of hazy images.