Image Haze Removal of Optimized Contrast Enhancement Based on GPU

Abstract

In the domains of computer vision and graphical computation, image haze removal has been a significant issue. By the use of haze removal process, it can significantly improve the visibility of the scene in the image. However, most of the haze removal algorithms bring high computational cost and make algorithms failed in processing huge amount of images. In this paper, we propose a parallel image haze remove algorithm, adopting optimized contrast enhancement approach, to optimize the performance based on GPU platform. The optimization from the proposed algorithm obtains performance acceleration with about 5 times as compared the original version while the haze removal effect is the same. Some haze free images and its original hazy images are shown in the later chapter during this paper. Our work after improvement can process a single picture in a much higher speed after optimization and make it more sufficiently fast for large-scale application which needs image haze removal in computer vision area.