Research On The Rapid Construction Method Of Complex Scene Based On Scene Graph

The progress of data acquisition and modeling technology contributes to the construction of smart city.The current data acquisition technology has been able to provide us with different kinds of products,including tilt model,point cloud model,urban BIM model and so on.At the same time,the rapid increase in the amount of urban 3D model data has also brought greater challenges to the organization and management of smart city data.In web applications,it takes time to transfer complex urban 3D data from the server to the client.So how to optimize the rendering efficiency and user experience of urban 3D scene by combining the spatial cognitive characteristics and user attention of urban 3D model under the viewpoint is very important.The rendering of urban 3D scene involves a large number of models,and the computer performance is limited.Arbitrarily storing all models in one folder will greatly reduce the data processing efficiency when calling the model.There are also problems of storage redundancy and semantic fragmentation on the storage boundary.In addition,the loading order of 3D model is not taken into account in the progressive transmission process,the entities concerned by users are not loaded first,and the real-time browsing effect is not good.In order to solve this problem,this paper designs a rapid construction method of complex scene based on scene graph,including the complex 3D scene data organization method combined with scene graph and the complex 3D scene data scheduling method driven by visual perception(combined with the geographical relationship provided by scene graph).Then a complex scene rapid construction system based on scene map is designed to practice the method proposed in this paper.The practical results show that this method can effectively support the smooth loading of large-scale urban scenes.The main research contents and research results of this paper are as follows:(1)Firstly,the research background and research significance of this paper are introduced,and the research status and development trend of complex 3D scene visualization at home and abroad are summarized.At the same time,the research purpose and research content of this paper are proposed.(2)The rendering of urban 3D scenes involves a large number of models,and the computer performance is limited.Arbitrarily storing all models in a folder will greatly reduce the data processing efficiency when calling models.There are also issues of storage redundancy and semantic fragmentation at storage boundaries.We propose a 3D scene management method based on Adaptive Quadtree and Scene Graph(AQT-SG)to solve the above problems.According to the spatial distribution characteristics of 3D scenes,this method uses adaptive quad-tree to organize macro-scale and meso-scale urban 3D scenes,traverses the quadtree from bottom to top,builds a flexible multi-scale 3D model,and uses the scene Figure organizes and manages 3D scenes at the microscopic scale.We validate the proposed method with park data from an intelligent park management system.Through the comparison of the results of the visualization of large and complex 3D scenes and the storage redundancy experiment results,it is verified that the method optimizes the data organization efficiency and improves the visual experience.(3)Aiming at the problems of low data scheduling efficiency and relatively delayed rendering when constructing complex urban 3D scenes,a visual perception-driven strategy based on scene graph is proposed.According to the spatial distribution characteristics of 3D scene data,this strategy uses scene graphs to organize local 3D scenes in cities.The algorithm adopts a detail simplification algorithm to simplify the 3D model of the city into four resolution levels.On this basis,a visual perception-driven strategy based on scene graph is designed.This strategy utilizes the good relationship properties between geographic entities provided by the scene graph to construct a visual perception evaluation model,and constrains the adaptive scheduling of the model at different levels of detail.Results of several experiments successfully represent the effectiveness of this method in improving data scheduling and accelerating local 3D scenes’ construction.(4)By developing a complex scene rapid construction system based on scene map,it is verified that the complex scene management method combined with scene map can support the rapid construction of urban 3D scene.