Research On Generalization Simplification And Classification Visualization Of 3D City Model Based On Cesium Platform

With the advancement of digital cities,the demand for urban planning and management has gradually increased,and the application fields of 3D urban models have become more and more extensive,including architectural visualization,light visibility analysis,urban flooding analysis,and flood simulation.However,the current development of smart cities has led to an increasing amount of data for refined 3D models,resulting in a drastic reduction in the 3D frame rate when loading batch models at one time during visualization.Online web-based visualization has become a bottleneck restricting its application development.And it is impossible to reveal the structural and dynamic evolution behind the complex forms of the city.Therefore,how to effectively simplify the complex 3D city model and realize the visualization of 3D city model classification is an urgent problem to be solved.In this context,this paper studies the standard data source(CityGML)in the field of 3D urban model,aims at comprehensively simplifying 3D city data and model classification,improves model loading efficiency and enhances classification visualization effect,and establishes massive mass based on Cesium.The 3D city model rapid classification visualization system,the specific research content is as follows:First,it is aimed at the comprehensive simplification of massive 3D urban models.This paper analyzes the limitations of the traditional 3D model simplification.On the basis of not changing the appearance of the model foundation,the new idea of extracting the neighboring edge search algorithm is proposed.Second,a three-dimensional urban model classification study based on deep learning.The traditional 3D model classification method is to classify the artificially designed feature descriptors,which will cause the lack of model information to a certain extent.Therefore,this paper combines the characteristics of 3D city model data,applies the deep learning method to the field of 3D city model classification,and improves the existing deep learning framework to achieve efficient and accurate classification of the model.Third,based on the Cesium city 3D model rapid classification visualization design and implementation.In the process of visual realization,the Bing map service is used as a two-dimensional base map,and the three-dimensional building model with the simplified and classified classification is equipped with the upper layer,and the threedimensional urban model is quickly rendered and classified visualized on the Cesium platform.Finally,through the comparison of the relevant experimental results,it is tested that the proposed model simplification method has a better simplification rate while preserving the visual effect of the original model.The improved model classification algorithm can enhance the visual appearance during visualization.Richness and diversity.The Cesium-based online 3D city model comprehensive simplification and classification visualization system can meet the requirements of rapid classification visualization.