基于边折叠的实景三维模型简化算法

Mesh Simplification Algorithm for Photorealistic 3D Models Based on Edge Collapse

三维模型轻量化是实景三维中国建设的当务之急,对三维模型进行一定简化是较为合适的途径。在模型简化算法中,二次误差测度QEM(Quadric Error Metrics)算法是较为经典的算法。但传统的QEM算法在简化过程中没有专门的机制来保护重要细节,并且简化后模型网格质量有待进一步优化。为此,本文提出了一种基于边折叠的实景三维模型简化与优化方法,该方法中的简化算法引入了顶点近似曲率及体积误差作为约束条件,以改变边折叠代价从而使三维模型在简化的同时能够保持模型的重要细节,达到较好的简化效果。在改变边折叠代价的同时加入边界保护条件,有效保护了模型边界。最后针对简化后的网格进行了拉普拉斯网格优化处理,用于三角形形状优化和特征保持。本文使用的数据源是采用倾斜摄影测量方式采集影像并使smart3D三维重建得到的格式为OSGB的实景三维网格模型数据,使用这一数据源进行实验研究,并和经典的QEM算法进行对比。研究结果表明,本文算法很好地保留了三维模型的细节特征,同时提高了简化模型的网格质量。该算法适用于实景三维模型的轻量化。

The lightweighting of three-dimensional models is a pressing need in the construction of realistic three-dimensional environments in China.Simplifying three-dimensional models to a certain extent is a suitable approach.In the realm of model simplification algorithms,the Quadric Error Metrics(QEM)algorithm is considered a classic method.However,traditional QEM algorithms lack a dedicated mechanism to preserve crucial details during the simplification process,and the post-simplification mesh quality needs further optimization.To address these challenges,this paper proposes a real-world three-dimensional model simplification algorithm based on edge collapsing.The aim is to better preserve key details and enhance the mesh quality of the simplified model in the process of three-dimensional model lightweighting.The simplification algorithm in this method introduces vertex approximation curvature and volume error as constraints,altering the edge collapsing cost to maintain important details while achieving effective simplification.By incorporating boundary preservation conditions while modifying edge collapsing costs,the algorithm effectively safeguards the model's boundaries.Finally,post-simplification mesh optimization is performed using the Laplacian mesh optimization method to optimize triangle shapes and preserve features.Experiments were conducted to verify the proposed algorithm and compare it with the traditional QEM algorithm.The research results indicate that our proposed method retains regions with rich details better than the traditional QEM algorithm,significantly reducing the number of elongated triangles and improving mesh quality.This algorithm is applicable to the lightweighting of real-world three-dimensional models,contributing to enhanced model performance and visualization effects.