基于暗原色先验理论,提出一种单幅图像去雾算法,并对其中的可调参数进行讨论,分析参数变化对去雾效果的影响。针对原算法中提到不用特殊处理的天空区域进行验证,发现此区域需要进行单独处理,通过设置阈值将天空区域隔离出来进行处理,并取得了较好的效果。对引入的导向滤波算法进行研究,分析算法中各项参数对算法实时性的影响。为了进一步提高算法实时性,对图像进行缩小处理,以减少求取透射率所需的时间,再利用插值法将透射率图还原至原图尺寸,从而得到无雾图像。实验证明,此方法在保证去雾效果的前提下,可将算法整体运算时间降低85.7%。

Based on the dark channel prior, we propose a dehazing algorithm for single image, discuss the adjustable parameters of the algorithm, and analyze the impact of parameter variation on the final dehazing results. In the primal algorithm, the sky area is mentioned as no need of special handling. However, we find that this area needs to be treated separately. We set threshold to isolate the sky area separately and receive better effect. Meanwhile, we study on the introduced guided-filter, and analyze the impact of parameters on the real-time performance of the algorithm. In order to improve the real-time performance of the algorithm, we reduce the picture firstly in order to shorten the calculation time, and then enlarge the image by image interpolation method to get haze-free image. The experimental results show that the operation time of the algorithm can be reduced by 85.7% with the premise of image haze removal.