摘要
针对现有地貌晕渲算法多是基于单核单线程编程模式从而导致计算效率较低的问题,提出利用多核并行计算模式对现有地貌晕渲算法进行并行化改进。首先,通过格网分割方式对原始数字高程模型(DEM)数据进行分割从而实现数据分块;其次,利用.NET环境下的Parallel类对分块数据进行并行晕渲处理,得到各个分块数据的晕渲结果;最后,对各个分块数据的晕渲图像进行拼接,从而得到完整的地貌晕渲图像。实验结果表明:并行化改进算法的计算效率明显高于现有单核单线程地貌晕渲算法,且参与计算核数与晕渲效率基本上呈线性增长关系;同时,分析发现地貌晕渲的立体真实感与光源参数的设置极其相关。
Most of the exiting hillshading algorithms are implemented based on single-core single-thread programming model, which makes them have lower computational efficiency. To solve this problem, an improved algorithm for parallelizing the existing hillshading algorithms based on multi-core programming model was proposed. Firstly, the original Digital Elevation Model （DEM） data were divided into several data blocks by grid segmentation. Secondly, these data blocks were shaded in parallel using the class Parallel under the . Net environment to generate shaded image of each block. Finally, the shaded images were spliced into a complete hiUshading image. The experimental results show that the calculation efficiency of the improved parallelized algorithm is obviously higher than that of the existing shading algorithms based on single-core single- thread programming, and there is an approximate linear growth relation between the number of the involved cores and the shading efficiency. Additionally, it is also found that the three dimensional and realistic effect of the hiUshading image is extremely relevant to the parameter setting of the light source.
出处
《计算机应用》
CSCD
北大核心
2017年第7期1911-1915,1920,共6页
journal of Computer Applications
基金
国家自然科学基金资助项目(41201381
41376108)
公路地质灾变预警空间信息技术湖南省工程实验室开放基金资助项目(kfj150605)~~
关键词
多核环境
数字高程模型
地貌晕渲
并行计算
数据分块
图像拼接
multi-core environment
Digital Elevation Model （DEM）
hillshading
parallel computing
data segmentation
image mosaic
