CAJ | 학술논문

本文在分析典型多边形栅格化算法的基础上，研究了串行算法并行化思路，提出一种多边形栅格化算法并行框架。该并行框架包括 MPI 与 OpenMP 的双层并行模式、顾及负载均衡的矢量多边形数据划分方法、多边形栅格化基本算子调用接口。利用本文形成的并行框架对扫描线算法、边界代数法进行了并行化，并利用大规模土地现状数据验证本文所提出的并行化方法的有效性。试验结果表明，该方法能够解决矢量多边形栅格化串行算法快速并行化的问题，并行化后的算法大大减少了矢量多边形转换时间，具有良好的并行效率。
본문재분석전형다변형책격화산법적기출상，연구료천행산법병행화사로，제출일충다변형책격화산법병행광가。해병행광가포괄 MPI 여 OpenMP 적쌍층병행모식、고급부재균형적시량다변형수거화분방법、다변형책격화기본산자조용접구。이용본문형성적병행광가대소묘선산법、변계대수법진행료병행화，병이용대규모토지현상수거험증본문소제출적병행화방법적유효성。시험결과표명，해방법능구해결시량다변형책격화천행산법쾌속병행화적문제，병행화후적산법대대감소료시량다변형전환시간，구유량호적병행효솔。
In this paper,the basic principles of representative rasterization algorithms are discussed to research on the methods for parallelizing sequential rasterization algorithms.As a result,a parallel framework intended for parallelizing rasterization algorithms is presented.The proposed parallel framework consists of a dual-level parallel pattern using MPI and OpenMP,a reasonable decomposition method for vector polygons and an interface for calling the sequential rasterization algorithms.The scanline method and boundary algebra filling method are parallelized using this approach and the land use data with great volume is used in the experiments.Some useful conclusions can be drawn from experimental results:(1 )the traditional rasterization algorithms can be parallelized rapidly using proposed parallel method;(2)the two parallel rasterization algorithms can dramatically decrease the conversion time and obtain good parallel efficiency.