三角网至规则格网的快速生成方法及其应用

摘要

Generating grids from triangulation takes an important role in applications ofterrain interpolation and pricewise rectification for remote sensing image registration.The key problem is to quickly and accurately search the triangle whereis each grid point, and then interpolate each grid point based on the triangle vertex information.Aim at efficientlyfinding grid points in the triangle positioning and time-consuming interpolation, this paper presents an adaptive scanning fill algorithm to quickly and accurately locate each grid point in the triangle.Due to the same operation in the application of terraininterpolation and image rectification, we use GPU technolo-gy to speedup the process.Ground points derived from automatic aerial triangulationprocessing are used to carry out a terrain interpola-tionexperiment, while a pair of large-size remote sensing image is used to perform a rectification experiment.The results show that the proposedmethod can provide a flexible way for terrain reconstruction and image rectification with high efficiency.%三角网至规则格网的生成在地形内插和遥感影像配准中小面元微分纠正等方面都具有重要应用。其关键在于快速准确地搜索每个格网点所在的三角形，然后根据三角网顶点信息内插每个格网点的信息。针对格网点在三角网中的定位和内插计算量大的问题，本文提出了一种自适应扫描线填充算法以快速准确地搜索定位出每个格网点所在的三角形；针对地形内插或者小面元微分纠正中每个三角形具有相同运算的特点，采用 GPU技术实现了TIN格网内插以及小面元微分纠正并行加速。分别采用大范围自动空三获取的三维点进行地形内插以及大尺寸遥感影像进行微分纠正实验，结果表明本文方法可以有效提高效率。