Modeling approach for railway alignment LoD mesh scene based on boundary projection and GPU render
The LoD(Level of Detail)grid terrain model is an important foundation for building a 3D visual scene of railway alignments.However,due to the complex dynamic characteristics of LoD grid models with multiple levels and scales,as well as the independent and blocky structures of classical LoD quadtree 3D grid terrain,many difficulties are posed for terrain clipping-trepanning.The alignment terrain integration when modeling railway alignments on a global digital terrain platform.To address these issues,this article proposed a trepanning method for railway alignments based on the outer boundary projection on global LoD grid terrain,as well as a dynamic combination and overlay method for railway alignments and LoD terrain using GPU vertex parsing,thereby achieving efficient modeling of 3D scenes of railway alignment LOD grid terrain.Specifically,this paper combined the GPU and CPU methods to filter the terrain clipping area to shield the redundant triangles inside the area and,hence,leaving the trepanning regions.The peripheral vertices were moved to the corrected position on the edge of the area through projection to complete edge filling,thereby avoiding the triangulation integration modeling on regular grid and achieving regular grid compatible expression of terrain.Subsequently,based on setting global unified texture coordinates for terrain vertices,different texture coordinate transformational relationships were established in GPU for terrain texture and overlay image to achieve seamless expression between them.For the interaction of overlay image,affine transformation was adopted to overcome the difficulty of multiple spatial nonlinear transformations and solve the asynchronous problem of texture coordinate conversion during interaction.The effectiveness of the method is verified through a section of 184 km railway alignment 3D modeling case.The results are shown as follows.The method proposed in this paper can clip and modify specific terrain areas.The use of GPU vertex parsing can achieve efficient integration of external railway models and real terrain.The rendering efficiency of large-scale 3D railway scene modeling can be increased from 40～70 frames to 55～77 frames,with an optimization rate of about 30%.
railway alignmentslevel of detail modelterrain cuttingimage overlapping3D visualization
国家科技期刊平台
NSTL
万方数据
