Design and implementation of virtual reality dynamic modeling
Based on the tetrahedral subdivision of point cloud data,an optimization method for virtual reality dynamic modeling in intelligent driving was proposed.The complete process involved the extraction of point cloud data,data enhancement,multi-layer convolution,multi-layer perception,intelligent decision support,data bridging and dynamic modeling.In the classification recognition stage,due to the sparse LiDAR point cloud data,data enhancement was made through the combination of Delaunay tetrahedral subdivision,rotation and scaling.The model training was finished by PointNet deep neural network and the results of the tests indicated that the accuracy could be increased to 90.6％.During the dynamic modeling phase,CAD models were adopted based on the previous classification recognition to assist modeling and Delaunay tetrahedral subdivision was utilized to generate three-dimensional models.The two phases were coupled by database.The feasibility of the proposed method was verified through dynamic modeling results.
data enhancementtetrahedral subdivisiondynamic modelingdatabase
万方数据
维普
