高速铁路线路护栏完整性自动检测系统研究

Automatic Detection System of Fence Completeness for High-speed Railway Line

高速铁路由于运营速度较快需要1个封闭的运行环境。本文提出1种基于机器视觉技术的高速铁路线路护栏完整性自动检测系统,以及针对高铁两类护栏的完整性识别方法。该系统安装在高速综合检测列车上,主要包括两台摄像机和图像处理与识别装置。两台摄像机分别拍摄两侧护栏,图像通过光纤传输到计算机,由计算机处理图像并识别图像中护栏的完整性。针对高速铁路广泛采用的水泥栏杆型护栏和声屏障型护栏,本文分别提出基于局部动态阈值分割以及结合动态轮廓模型和高斯混合模型的完整性识别算法。本系统在高速铁路现场进行了实验,实验结果表明:本文提出的系统和识别算法可以有效识别护栏缺失和破损,并对不同天气、光照等环境因素有较好的适应性。

High-speed trains need isolated running environments. In this paper, the automatic detection system of fence completeness for high-speed railways was presented. This system used two cameras installed on the high-speed CIT （Composited Inspection Train） to shoot the fences on both sides,and then transmitted the images by optic fibers to a computer,which processesd the images and recognized the completeness of the fences automatically. Two different algorithms were presented for two different types of fences wildly used along China high-speed railway lines：the cement balustrade and acoustic shield. For cement balustrades, the two-stage method mainly based on local dynamic thresholding was adopted; and for acoustic shields, the detection algorithm based on the active contour models and mixed Gaussian model was presented. Field tests were conducted along the High-speed Beijing-Shanghai Railway Line. The test results show that the proposed system and algorithm can detect the losses and breakages of the fences effectively and suit the different weather and illumination environmental conditions.