Fast Dark Channel Prior Based Haze Removal from a Single Image

Abstract

Outdoor images may suffer from haze, and the clearness of the image is greatly lost. The haze removal, which is known as dehazing, is an important issue. In this paper, we propose a fast haze removal method, which is an improving version of an existing method using dark channel prior proposed by He et al. [1]. In our proposed method, we significantly reduce the computational time by improving the method to estimate the dark channel. Our proposed dark channel estimation method uses a down-sampled image and do not need a soft-matting process [2]. Experiments with haze images show that our proposed method is faster and an acceptable quality level compared with the existing He's dark channel method.