摘要
矢量数据化简是制图综合领域的重要研究内容,但是传统的简化算法,如道格拉斯-普克(Douglas-Reucker,DP)算法等,多基于串行计算方式,存在计算量大、执行效率低的问题。本文利用共享内存并行处理技术对DP算法进行并行化改造,提出了点级并行、要素级并行以及双并行三种并行策略,并进行了相关实验验证与对比分析。实验结果表明,在合适的线程数量下,三种并行策略均可提高矢量数据的化简效率,其中双并行算法可以最大限度地利用计算机的多核计算资源,取得最佳的加速比。
The simplifcation of vector data is an important research content in cartographic generalization.However,tradi-tional algorithms such as Douglas-Peucker(DP)algorithm are mostly based on serial computing,which are restricted by the large amount of calculation and low efficiency.In this paper,the DP algorithm is improved by using the shared memory parallel processing technology OpenMP,and three strategies are proposed to accelerate the simplification process,namely the point-level parallel,feature-level parallel,and double parallel strategy.Experiments and comparative analyses are carried out to verify the validity of these strategies.The experimental results show that all proposed parallel strategies can significantly improve the effi-ciency of vector data simplification with a proper number of threads,and the double parallel strategy can maximize the use of mul-ticore resources and achieve the best speedup ratio.
作者
黄飞龙
张吉才
宋默
郭力刚
HUANG Feilong;ZHANG Jicai;SONG Mo;GUO Ligang(Unit 61618,Bejing 100088,China;Nankai University,Tianjin 300350,China;Unit 61206,Bejjing,100042,China)
出处
《测绘科学与工程》
2019年第3期61-66,共6页
Geomatics Science and Engineering
关键词
矢量数据化简
DP算法
共享内存
并行处理
OPENMP
vector simplification
Douglas-Peucker algorithm
shared memory
parallel processing
OpenMP
