基于自适应指数加权移动平均滤波的快速去雾算法

经典的暗原色先验去雾算法易丢失图像细节信息,基于保边滤波的去雾算法虽可以有效保护图像细节,却耗时较长。针对以上问题,提出一种能够很好地保护图像边缘细节且耗时较短的自适应指数加权移动平均滤波算法,并与改进的暗通道结合,实现快速去雾。首先,对暗通道加以改进并求得透射率粗分布;再利用自适应指数加权移动平均滤波算法对透射率进行优化;之后修复明亮区域透射率,避免颜色失真;最后通过变换大气散射模型求解得到去雾图像。实验结果表明:本文算法具有很快的执行速度,且经本文算法处理后的去雾图像质量较高,在有效边缘强度、色彩还原能力、结构信息这三个无参考客观评价指标下均表现不错。

Fast Defogging Algorithm Based on Adaptive Exponentially Weighted Moving Average Filtering

The classical dark channel prior defogging algorithm easily loses image details.Alternatively,the defogging algorithm based on edge-preserving filtering can effectively protect image details;however,it is time-consuming.Aiming at the aforementioned problems,this paper proposed an adaptive exponentially weighted moving average filtering algorithm that protected image edge details while taking less time.Combined with an improved dark channel,this method achieved fast and precise defogging.First,the improved dark-channel algorithm was applied to obtaining a rough distribution of atmospheric transmittance.Second,the transmittance was optimized by employing the adaptive exponentially weighted moving average filtering algorithm.Subsequently,the transmittance of the bright region was repaired to avoid color distortion.Finally,the defogged image was processed using the transformation of the atmospheric scattering model.The experimental results show that the proposed algorithm has a high execution speed;moreover,the defogged image processed using the proposed algorithm has good performance under the following three nonreference objective evaluation indexes:effective edge intensity,color reproduction ability,and structural information.