Research On Automatic 3D Modeling Technology Of Traffic Scene

In recent years,automatic 3D modeling of large-scale scenes has been a hot topic in computer graphics.In the micro-traffic simulation,the introduction of three-dimensional urban traffic scene not only provides more visual information for decision-making and a more intuitive basis for judgments,but also improves the capability of fine-grained simulation.However,for large-scale three-dimensional urban traffic scene,there are issues including data diversity,complex structure,manual construction workload,low efficiency,So it is important to automatically build large-scale three-dimensional traffic scene.Therefore,based on SUMO simulation platform and OpenSceneGraph 3D platform engine,this paper proposes an algorithm to automatically build large-scale three-dimensional urban traffic scene.The main research contents include:1.This paper constructs the 3D traffic scene by modeling the geometric data,and puts forward the framework design that fusions the SUMO simulation platform and the OpenSceneGraph.A TinyXML parse tree is introduced for the OpenStreet Map data.The tool parsed the control point data and vehicle simulation route data from the map data.And the conversion of the data from latitude and longitude coordinates to plane coordinates is realized by Gaussian projection.2.This paper proposes an algorithm to automatically construct the planar road structure by automatically identifying the road attribute structure and calculating the deflection angle of the rectangular bounding box.The algorithm can automatically construct the three-dimensional planar road structure with complex configuration.For the complex structure such as intersection,this paper introduces the retraction coefficient ? on the basis of the Bezier curve,and realizes the automatic construction of the complex intersection geometry.3.In this paper,the automatic identification of the position of the three-dimensional cross-structure is realized by judging the road space relationship,and the method of three-dimensional traffic geometric topology is automatically identified by the method of tangent and the tangent of the projection point in the Z-axis deflection angle.And the AABB collision detection algorithm is introduced to realize the optimization of the time complexity of the decision method of the road space relationship from 2O(N)to O(log N).4.In this paper,the cubic spline interpolations algorithm is used to smooth the constructed road model,and the large-curvature three-dimensional space structure is built by introducing the clothoid curve and the Frenet-Serret model.Realistic rendering of 3D scene is realized by Blinn-Phong illumination model and automatic texture mapping algorithm.At the same time,this paper proposes an adaptive matching algorithm of 3D model unit and scene model to realize the stereo vision optimization of three-dimensional scene.The paper proposes a large-scale three-dimensional traffic scene automatic construction algorithm,and through the algorithm to realize the automatic recognition of the planar road structure and the cross structure and the automatic construction of the 3D scene.Three-dimensional scene is also optimized by the smoothing technique and the scene realistic rendering technique.The study can not only build large-scale three-dimensional traffic scene based on real road data,but also provide the interface of three-dimensional traffic simulation and geometric scene interaction.It also provides a platform to verify the effect of traffic experiment and traffic control,and increasing the flexibility and diversity of 3D scene construction.