为了分析并提高在多核计算环境下大数据量激光雷达数据内插DEM效率问题,利用多线程技术,设计了在多核计算环境下DEM分块内插并行算法,讨论了并行算法的流程和技术要点,并实验分析了分块大小和搜索半径对并行算法的加速比影响。实验表明:在分块较小时,加速比与核数具有线性增长关系;而内插搜索半径越大,加速比越小,但影响不十分显著。总体上,多核并行算法一定程度上提高了DEM内插效率,但是会受到计算机内存速度的限制。
In order to analyze and improve the efficiency of DEM interpolation of large scale 3D laser scanning dataunder multi-core computing environment,the authors proposed aparallel DEM interpolation algorithm combinedwith multi-thread technology.The procedure and technical key points of the parallel interpolation algorithm werediscussed.And the influence of sub-block size and search radius on process speed was investigated by real experi-ments.Experiments showed accelerate rate has alinear growth relation with the number of CPU cores when sub-block size is less.The increasing of search neighborhood distance will cause aslightly decreasing influence to the ac-celerate rate.Generally,the multi-core parallel algorithm could improve the efficiency of DEM interpolation but willbe constrained by the speed of computer memory.