基于滤波的单幅图像去雾方法

基于Tarel方法提出一种新的基于滤波的单幅图像去雾方法.先利用双边滤波求出初始的大气散射光,较好地保持了边缘平滑纹理;再利用自适应中值滤波器进一步求得边缘清晰、纹理平滑、景深信息真实的大气散射光;然后基于分层搜索的四叉树分解方法求得大气光;最后,依据雾天退化模型得到复原的图像.与经典去雾算法对比结果表明,使用该方法复原的图像更接近真实图像,特别对于纹理较丰富的区域和远景区域去雾效果更明显.

Single Image Dehazing Method Based on Filters

This paper introduces a novel single image dehazing method based on filters. Firstly, in order to maintain a good edge smooth texture, the initial atmospheric scattering light is estimated via bilateral filtering. Then, the initial atmospheric scattering light is refined via the adaptive median filter to keep sharp edges, smooth texture, and actual depth information. Next, the atmospheric light is obtained by the hierarchical searching method of quad tree subdivision. Finally, the dehazed image is restored by solving the atmosphere attenuation model. Compared with the results obtained by several existing classical dehazing methods, the experimental results show that the dehazed image by our proposed method is closer to the real scene image. Especially, it exhibits obvious superiority in dehazing regions with rich texture and distant scene.