高速铁路无砟轨道表面裂缝三维图像自动识别算法

Automatic Detection Method for High-speed Railway Ballastless Track Surface Cracks Based on 3D Image Technology

无砟轨道表面伤损的自动检测技术是当前高速铁路检测与监测的关键技术。采用三维图像技术,将原始三维图像转换为二进制图,基于三维光影模型的轨道结构表面裂缝的三维图像识别算法,采用连通域分析与线性形态分析方法,了轨道结构裂缝识别中图像噪声消除算法,从而提高轨道结构表面裂缝自动识别的准确率。室内试验对比结果表明:课题组研发的高速铁路轨道表面伤损检测系统,可获得高精度的裂缝长度、宽度以及深度的数据信息,轨道板裂缝最大宽度的识别结果相对误差为6.25%、9.68%,裂缝长度的测试识别相对误差为1.39%、2.92%,平均深度的测试识别相对误差为15.69%、13.04%。采用提出的裂缝识别算法可实现100%准确率的裂缝自动识别。

Automatic detection technology for the surface damage of ballastless track has become one of the key technologies for high-speed railway maintenance management.The 3 D image technology was used to convert the original 3 D image into the binary one.The 3 D image recognition algorithm based on the 3 D shadow model was established for the surface crack detection of the track structure.The connected domain analysis and linear morphology analysis method were used to establish the image noise elimination algorithm for the track structure crack identification,so as to improve the accuracy of automatic surface crack identification.According to the model test results,the data of crack length,width and depth are accurate,using the system built by the research group in the laboratory.The relative error of the identification results of the maximum width of the track surface crack is about 6.25%and 9.68%.The relative error of the identification results of the crack length is about 1.39%and 2.92%.The average depth test identification relative error is about 15.69%and 13.04%.The proposed track structure surface crack detection method can achieve automatic identification with 100%accuracy rate.