The Methods Of Extracting Railway Cross Section’s Terrain Surface Attributes Information

As a new technology combining both the three-dimensional positioning and aerial photogrammetry, LiDAR has been widely applied in the railway survey. It is an important application of LiDAR technology in the railway survey to acquire the high accuracy DEM which can be available by classifying3D point cloud data that is acquired by LiDAR technology and the cross-section of the railway based on LiDAR.Because of the lack of relevant texture information, LiDAR data cannot provide cross-section with the corresponding attribute information.However, the attributed information of the rail cross-section can be used as a priori knowledge for the railway’s line selection and design.Given that the LiDAR system also obtained high-resolutionaerial images using digital equipment during its flight. The high-resolution aerial images contain rich surface features. Many scholars have also been using object-oriented approach to extract surface features from the high resolution image. Under this background, in order to realize the abtracting and the assigning the property of the cross-section of railway, this article did the research on the method of the auxiliary aerial images of LiDAR data for terrain classification.The main contents and results of this study are as follows:1. Analyzing the mathematical model and the technical requirements of the railway’s plane line in building the railway line object. Based on the constructing the Triangulated Irregular Network of the discrete points, introducing the way of abstracting the cross section of railway using the Q, function to judge the topological relations between the points and triangular.2. Studying the LiDAR auxiliary aerial images of the object-oriented multi-scale segmentation techniques. Analyzing the effect of the "over-segmentation" phenomenon through the edge detection on the image by Canny operator: Under the method of segmentation scale brute-force, building the "Multi-factor of Segments’ Quantitative Indicators" as a segmentation evaluation to determine surface features optimal scale, and use indirect precision evaluation method to evaluate the results of the partition scale from the function, Finally, make the terrain classification by combining the elevation, spectral, and texture of the surface features.3. Designing the secondary development of the three-dimensional terrain scenes system based on the Skyline and ArcEngine.which is combing of the abstracting remote sensing information of the oriented-object technology and engineering applications to get the terrain information, lead the worker to find the field administration boundary quickly. And realizing some function of the software.4. Summarizing the workflow of the object-oriented classification by high-resolution aerial images assisted by LiDAR. and evaluating the experimental results by the user accuracy, producer accuracy, the global accuracy and Kappa coefficients.It is issued that the aerial images combining the LiDAR data processes the object-oriented of terrain classification not only simplifies the classification procedure, but also improves the user accuracy, producer accuracy, overall accuracy and Kappa coefficient compaired by using only the aerial images to do the object-oriented remote sensing classification. Meanwhile, it is meanful to combine the object-oriented classification of remote sensing imformation and the engineering applications in conjunction with the correction and modification manually to improve the efficiency of the work of the cross-sectional topography change point property assignment. The result of this paper has a certain reference value of the further development of the related work.