A fast method for single image haze removal based on multiscale dark channel prior

Abstract

Aiming at the problem that the computational process in the method for image haze removal based on the dark channel prior is too complicated and too time-consuming, we propose a fast method for single image haze removal based on multiscale dark channel prior. In order to solve the problem that it takes too much time on a high resolution image, we choose to optimize the image haze removal based on dark channel prior algorithm by the method of multiple scales. We deal with rough resolution images and use fast minimum filtering and fast guided filtering to speed up the haze removal algorithm. Therefore, the speed of calculation is accelerated while maintaining the good image effect after haze removal.