一种快速的单幅图像去雾算法

A Fast Effective Single Image Dehazing Algorithm

为了解决基于暗原色先验的单幅图像去雾算法运行效率低,以及去雾后在图像的明亮灰白区域存在图像色彩失真的问题,提出一种快速有效的单幅图像去雾算法.该算法基于HSI色彩空间进行粗略的透射率以及大气光的估计;然后,用导向滤波对粗略的透射率进行平滑,再利用阈值法对灰白明亮区域的透射率进行修正,得到最终的透射率;最后,进行色彩调整得到复原图像.实验结果表明,本文算法具有很高的运行效率,能有效提高复原图像的清晰度和对比度.

A fast and effective single image dehazing algorithm is proposed in order to solve the problems of lowefficiency and the color distortion of the bright grey areas after dehazing algorithm of using dark channel prior. First,the algorithm estimates the rough transmittance and the atmospheric light based on HSI color space. Then,the final transmittance is generated after smoothing the rough transmittance by guided filter and correcting the transmittance of the bright grey areas by a threshold method. Finally,tone mapping are used to get the restored image. Experiments showthat the proposed algorithm has high efficiency,which can effectively improve the visibility and contrast of the restored image.