基于透射率和单目深度的团雾检测

Fog visibility measurement based on transmittance and mo-nocular depth

团雾导致的能见度剧变是目前高速公路交通安全的主要威胁之一,为了解决高速公路上团雾检测困难问题,提出一种基于透射率和单目深度的视频图像团雾检测算法。首先通过canny算子获取轮廓,以天空区域的像素均值作为大气光值,结合动态权重系数提升透射率值准确性;其次通过透射率和单目摄像机线性模型计算场景深度;最后根据能见度检测原理求解大气消光系数,获取该图像的能见度级别,实现对团雾的实时检测。实验分析表明,算法的准确率和及时性较高,能够满足实际场景需求。

The dramatic change of visibility caused by clump fog is one of the main threats to highway traffic safety at present.In order to solve the difficult problem of clump fog detection on highways,a video image clump fog detection algorithm based on transmittance and monocular depth is proposed.Firstly,the contour is obtained by canny operator,and the mean value of pixels in the sky region is used as the atmospheric light value,combined with dynamic weighting coefficients to improve the accuracy of transmittance value.Then we calculate the scene depth by transmittance and monocular camera linear model,and finally solve the atmos-pheric extinction coefficient according to the principle of visibility detection,and obtain the visibility level of the set image to achieve the real-time detection of clump fog.The experimental analysis shows that the ac-curacy and timeliness of the algorithm are high and can meet the actual scene requirements.