Efficient airlight estimation for defogging

This paper presents a new technique to efficiently estimate atmospheric light of visibility restoration algorithms from a hardware implementation perspective. The quality of defogged images depends on the accuracy of estimated atmospheric light, where it is typically estimated from the brightest portion of foggy images. In most existing algorithms, long execution time and large memory resources are required to find the largest intensity values on hardware platform. We explore an efficient method to measure atmospheric light by detecting sky region.