暗原先验算法在天空区域的处理与改进

暗原先验算法在天空区域去雾过程中会造成噪声过度放大和晕环现象，针对这个问题提出了依据天空隶属系数调整天空区域透射率的方法。求得有雾图像相应位置天空隶属度系数，自适应改变透射率调整幅度，并利用二次函数特性，对隶属度小值区域进行压缩，减弱近景透射率有效区域的调整，最后通过大气散射模型得到复原图像。大量实验结果表明，该方法得到的复原图像清晰自然，特别是在天空区域，保留部分浓雾区域去雾程度的同时抑制了天空区域的噪声和晕环现象，取得了很好的去雾效果。相对暗原先验算法，改进算法在天空区域得到更加准确的透射率估计，复原图像效果更佳。

Processing and improvement of the dark channel prior in the sky region

The method based on dark channel prior will cause excessive noise and halo phenomenon in the region of sky . To solve this problem ,a method of adjusting the transmission rate of the sky region is proposed based on the credibility of the sky .To obtain the credibility coefficient of the corresponding position of the fog image ,Changing the adjustment range of transmissivity adaptively ,and using characteristic of the quadratic function to change the map of the credibility coefficient ,then the haze‐free image can be recovered by the atmosphere scattering model .A lot of experimental results show that the recovered image by the proposed method is clear and natural ,especially in the sky region .Retaining the degree of haze remove in dense fog area and suppressing the noise and halo phenomenon in the region of sky .The improved method proposed in this paper can obtain better effect than the method based on dark channel prior in the processing of haze removal .