融合纹理信息的实景三维模型简化算法  

Mesh simplified algorithm considering texture information

作　　者：李鹏[1] 颜青松 曲英杰 邓非[1] LI Peng;YAN Qingsong;QU Yingjie;DENG Fei(School of Geodesy and Geomatics,Wuhan University,Wuhan 430079,China)

机构地区：[1]武汉大学测绘学院,武汉430079

出　　处：《测绘科学》2021年第10期151-158,166,共9页Science of Surveying and Mapping

基　　金：四川省科技计划资助项目(2019YFG0460)。

摘　　要：针对二次误差测度(QEM)网格简化算法难以保留局部细节特征的缺点,该文提出了一种融合纹理信息的网格简化算法。该算法通过高斯滤波和拉普拉斯运算对纹理贴图预处理来获取像素点的纹理高低值,并以此改变边折叠代价来优化简化效果,从而有效保留三维网格模型的细节特征,助于多层次细节特征等后续操作。该文依据倾斜摄影获取影像构网生成的三维网格数据为试验数据,采用该文算法进行试验,并与传统QEM算法比较。结果表明,该文算法简化效率及简化精度均高于QEM算法。Aiming at the shortcomings of the quadratic error measurement(QEM)grid simplification algorithm that it is difficult to retain the features of local details, a grid simplification algorithm fusing texture information is proposed. The algorithm uses Gaussian filtering and Laplacian operation to preprocess the texture map to obtain the texture height value of the pixels, and thereby changes the edge folding cost to optimize the simplification effect, so as to effectively retain the detailed features of the 3 D mesh model and help multi-level detail features and other follow-up operations. The article used the three-dimensional grid data generated by the oblique photography to obtain the image network as test data, and used the algorithm to carry out the experiment and compared it with the traditional QEM algorithm. The results showed that the simplification efficiency and simplification accuracy of the algorithm in this paper were higher than the QEM algorithm.

关 键 词：网格简化 高斯滤波 拉普拉斯运算 细节特征 

分 类 号：P208[天文地球—地图制图学与地理信息工程]