一种基于并行计算熵迁移策略的多分辨DOM数据生成算法

数字正射影像图(digital orthophoto map,DOM)数据通常以多分辨数据形式组织,并以切片的方式存储,而海量多分辨DOM数据的生成需要大量计算和大容量存储.针对此问题,提出一种基于并行计算熵迁移策略的并行多分辨DOM数据生成算法,以减少海量多分辨DOM数据的生成时间.该算法采用并行计算熵来衡量并行计算机系统的负载平衡程度,并以此判断何时需要进行负载迁移以及如何迁移.仿真实验表明,与串行算法相比,该算法能有效减少程序执行时间,并且能获得较高的加速比和并行效率.

An algorithm based on emigrating strategy of parallel computing entropy for DOM

Digital orthoimage map(DOM) data are usually organized in the multi-resolution data format and stored in slices.However,massive multi-resolution DOM data generation requires heavy computation and large-capacity storage.A parallel and multi-resolving DOM data generation algorithm was proposed to reduce the generation time of massive multi-resolution DOM data based on the parallel computing strategy of entropy migration.The algorithm employed the entropy not only to measure the load balance level among the nodes in a parallel computing system,but also to determine when and how to emigrate a load.Simulation results show that,compared with sequential algorithms,the parallel algorithm can effectively reduce execution time and achieve higher speedup ratio and parallel computing efficiency.