Research On Railway Environment Recognition Technology Based On UAV Image

Acquiring railway line environment data through unmanned aerial camera or lidar,which is real-time,continuous,fast and rich in spatial information,is a research hotspot in the field of railway survey.However,in the process of data post-processing for solid modeling,the large amount of image or graphic data obtained has caused great restrictions on the professional application of railway lines.Extracting forward line design information from rich environmental data of railway lines is the focus of new technical specialization application,among which fast encapsulation and identification of local rail data is the difficult point.Therefore,under the precondition of ensuring the accuracy and speed of the modeling,lightweight preprocessing of rail point cloud data is of practical significance to the environmental identification of the entire railway line.In this paper,based on the general post-processing method of point cloud for railway lines,the secondary development of point cloud for railway lines is carried out on the basis of Windows operating system and PCL point cloud technology improvement algorithm.The automatic partition of rail point cloud under railway line environment is achieved;furthermore,a fast encapsulation algorithm for rail point cloud is designed based on the geometric characteristics of rail section and line characteristics.The preprocessed line point cloud model can quickly identify rails and lightweight the track model,provided that the data accuracy meets the needs of the project.For the above research work,the specific work done in this paper is as follows:(1)Segmentation of the continuous data of the railway line study area is based on image and graphic data sources,respectively.Based on the analysis and analysis of the practical effect of the common split method and the design of the rail point cloud data segmentation algorithm combined with PCL point cloud database,the rail point cloud data in the threshold interval can be quickly and automatically segmented from the overall railway line data.(2)Based on the vs2017+PCL open source platform and combined with the geometric characteristics of rail cross-section,a method of rail point cloud data encapsulation based on the characteristics of rail cross-section is proposed by analyzing the common encapsulation method of rail point cloud data.(3)Validation of the algorithm for a project example.The iFly D1 unmanned aerial vehicle mounts the iCam Q5 mini tilt photogrammetry system to obtain image data of a section of the line area in Dunger and encapsulate the rail model.The point cloud of the line is preprocessed by the segmentation and encapsulation algorithm,which enables the fast modeling of the rail entity and meets the engineering requirements with accuracy.The results of an example show that the fast segmentation and encapsulation algorithm for rail point cloud is effective.This method can be used to identify long-distance rail point cloud data based on UAV image.