基于导向滤波优化暗原色先验的公路能见度检测方法

针对中国交通领域的能见度实时监测问题,通过复用公路沿线密集布设的监控摄像头,提出了一种以暗原色先验理论为基础的公路能见度检测方法。该方法针对析取的监控视频帧,通过改进的暗原色先验图像处理方法获取视频帧中特定位置的空气透射率,然后按照摄像头架设的几何参数及大气光学理论推导相应的道路能见距离。与传统的暗原色先验方法不同,现采用导向滤波对暗原色先验的空气透射率计算进行优化,可以获得更加符合真实图像的透射率结果。实验分析结果表明,该能见度计算方法可行,能见度检测结果具有较高的准确度,因此可用于解决公路能见度检测和交通安全预警等问题。

Road Visibility Measurement Based on Guided Filtering Optimized Dark Channel Prior

With regard to the real-time monitoring of roadside visibility in transportation, by reusing the densely distributed video cameras alongside the road systems, a roadside visibility measurement based on the theory of dark channel prior was approached. In this measurement, taking excerpted video snapshots as the input image, an extended dark channel prior-based image processing methodology is applied to assess the atmospheric transmission rate of given spot in the image, and then the corresponding roadside visibility distance is calculated according to the geometry parameters of cameras and atmospheric optics theory. In addition, different to traditional dark channel prior-based methods, guided filtering method is used to optimize the calculating process, thereby resulting in a better atmospheric transmission rate value. Experimental analysis indicates that our method is feasible and the computing results are more accurate, and hence is applicable in the field of roadside visibility monitoring and pre-warning in China.