基于天空区域分割的输电通道图像暗原色去雾方法

Dark Channel Prior Dehazing Method for TransmissionChannel Image Based on Sky Region Segmentation

输电通道的实时监测系统会受到恶劣天气情况的影响,尤其是北方冬季有雾天气较频繁,因此,去雾处理是实时监测系统的重要预处理环节。针对图片中存在大量天空区域或白色物体会导致原方法失效的问题,在暗原色先验方法的基础上开展研究,提出了基于天空区域分割的改进方法,获取了优化后的大气光值和透射率值,从而实现了有雾天气下的有效监测。经过对不同去雾方法进行研究和对比,实验结果表明基于天空区域分割的暗原色去雾方法能取得更好的效果,分别在全参考和无参考的图像质量评价指标上取得更高的水平,尤其是针对包含天空区域的输电线路图像,能够较好地进行图像预处理,并且在雪天情况下对比原去雾方法有非常明显的去雾效果。

The real-time monitoring system of transmission channels will be threatened in poor weather.Considering the frequent foggy days in northern winter,it is essential to dehazing in the preprocessing link.Since massive sky regions or white objects in the picture will invalidate the original method,this paper proposes an improved method based on the dark channel prior.We obtain the optimized atmospheric light value and transmittance value to realize effective monitoring in foggy days.Through researching and comparing different dehazing methods,the experimental results show that the proposed method presents better image quality evaluation index with full reference and no reference respectively.It is excellent in the preprocessing of transmission line images containing sky area.In snowy days,it features ideal fog removal compared with the original dehazing method.