基于暗通道先验的图像去雾方法

Image Defogging Method Based on Dark Channel Prior

针对采用暗通道先验还原图像时引起分界区域出现块效应的现象,采用软抠图对透射率进行优化处理,提出一种改进的暗通道先验图像去雾方法。基于大津阈值法分割有雾图像中的非天空与天空区域,然后估算天空区域的大气光值,通过梯度引导滤波法细化非天空区域的初始透射率,进而复原得到清晰的去雾图像。实验表明,该方法有效解决了暗通道去雾模型存在的大气光值估算不准、天空区域颜色失真等问题,能够获取更为清晰的去雾图像。

Aiming at the obvious blocking phenomenon in the edge region where the scene depth changes in the dark channel prior restoration image,soft matting algorithm is adopted to optimize the transmittance,and an image defogging method based on dark channel prior improvement is proposed.The Otsu algorithm is used to divide the sky and non-sky areas in the outdoor image,the atmospheric light value is reasonably estimated by using the accurately divided sky area,and the initial transmission rate is refined by gradient guide filtering in the non-sky area,and the clear demisting image is obtained by the atmospheric scattering model.Experiments show that the method,on the basis of retaining the original dark channel defogging method,effectively solves the problems of inaccurate estimate of atmospheric light value,color distortion in the sky area,and obtains a clearer defogging image.