Premium
A high accuracy surface modeling method based on GPU accelerated multi‐grid method
Author(s) -
Yan Changqing,
Liu Jimin,
Zhao Gang,
Chen Chuanfa,
Yue Tianxiang
Publication year - 2016
Publication title -
transactions in gis
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.721
H-Index - 63
eISSN - 1467-9671
pISSN - 1361-1682
DOI - 10.1111/tgis.12224
Subject(s) - computer science , grid , computational science , acceleration , kriging , interpolation (computer graphics) , cuda , spline (mechanical) , surface (topology) , algorithm , parallel computing , artificial intelligence , mathematics , image (mathematics) , physics , machine learning , classical mechanics , thermodynamics , geometry
A high accuracy surface modeling method (HASM) has been developed to provide a solution to many surface modeling problems such as DEM construction, surface estimation and spatial prediction. Although HASM is able to model surfaces with a higher accuracy, its low computing speed limits its popularity in constructing large scale surfaces. Hence, the research described in this article aims to improve the computing efficiency of HASM with a graphic processor unit (GPU) accelerated multi‐grid method (HASM‐GMG). HASM‐GMG was tested with two types of surfaces: a Gauss synthetic surface and a real‐world example. Results indicate that HASM‐GMG can gain significant speedups compared with CPU‐based HASM without acceleration on GPU. Moreover, both the accuracy and speed of HASM‐GMG are superior to the classical interpolation methods including Kriging, Spline and IDW.