为了解决自然灾害遥感影像中局部存在浓雾的问题,提出了一种基于暗原色先验和雾天图像退化模型的图像去雾方法,并利用导向滤波实现光学遥感影像的去雾.首先,结合自然灾害遥感图像的特点,利用阈值将图像分为浓雾区域和薄雾区域,采用不同的方法得出两个区域的暗通道图;然后,结合导向滤波对透射率图进行优化,再对图像进行对比度拉伸,提高图像的动态范围,并选取多幅遥感图像进行去雾试验.最后,设计了一组图像增强质量评价指标,对去雾结果进行定量分析.结果表明,该方法能从物理特性上明显去除雾的干扰,提高图像清晰度,增强图像色彩和细节,从而复原得到高质量图像,在一定程度上能满足自然灾害遥感图像去雾的要求.

In order to remove the haze in natural disaster remote sensing images,an approach based on dark channel and hazy image degradation model was presented,which could remove the haze by using guided filter.Firstly,images of natural disaster were divided into fog area and mist area by threshd,and different methods were used to obtain the dark channel of the images.Then,guided filter was used to optimize the transmmission map,and the contrast of the images was stretched to improve the dynamic range of the images.A series of natural disaster remote sensing images were chosen to test the alogrithm of haze removal.Finally,a series of evaluate parameters were proposed to assess the method.The result shows that the proposed algorithm can remove the haze of the images,improve the image quality,and enhance the color and detail of the images,so that the high-quality images can be obtained,which meet the requirements of haze removal for natural disaster remote sensing images to some extent.