A Hybrid Parallel Spatial Interpolation Algorithm for Massive LiDAR Point Clouds on Heterogeneous CPU-GPU Systems

Abstract

Nowadays, heterogeneous CPU-GPU systems have become ubiquitous, but current parallel spatial interpolation (SI) algorithms exploit only one type of processing unit, and thus result in a waste of parallel resources. To address this problem, a hybrid parallel SI algorithm based on a thin plate spline is proposed to integrate both the CPU and GPU to further accelerate the processing of massive LiDAR point clouds. A simple yet powerful parallel framework is designed to enable simultaneous CPU-GPU interpolation, and a fast online training method is then presented to estimate the optimal decomposition granularity so that both types of processing units can run at maximum speed. Based on the optimal granularity, massive point clouds are continuously partitioned into a collection of discrete blocks in a data processing flow. A heterogeneous dynamic scheduler based on the greedy policy is also proposed to achieve better workload balancing. Experimental results demonstrate that the computing power of the CPU and GPU is fully utilized under conditions of optimal granularity, and the hybrid parallel SI algorithm achieves a significant performance boost when compared with the CPU-only and GPU-only algorithms. For example, the hybrid algorithm achieved a speedup of 20.2 on one of the experimental point clouds, while the corresponding speedups of using a CPU or a GPU alone were 8.7 and 12.6, respectively. The interpolation time was reduced by about 12% when using the proposed scheduler, in comparison with other common scheduling strategies.

Peer Review #2 of "Comparative investigation of parallel spatial interpolation algorithms for building large-scale digital elevation models (v0.1)"

“…Manuscript to be reviewed Computer Science analysis (Zhang et al, 2017), urban growth simulation (Guan et al, 2016), Delaunay Triangulation (DT) for GIS (Coll and Guerrieri, 2017), spatial interpolation (Wang et al, 2017;Cheng, 2013;Mei, 2014;Mei et al, 2017;Mei, 2014;Ding et al, 2018b), and image processing (Wasza et al, 2011;Lei et al, 2011;Yin et al, 2014;Wu et al, 2018).…”

Parallelizing Multiple Flow Accumulation Algorithm using CUDA and OpenACC

“…They demonstrated performance improvements over a serial version on a CPU through a series of tests, and achieved speedup from 10 to 100. Wang et al [10] proposed a hybrid parallel spatial interpolation algorithm executed on both the CPU and GPU to increase the performance of massive LiDAR point clouds processing. Their experimental results have shown that their hybrid CPU-GPU solution outperforms the CPU-only and GPU-only implementations.…”

Big spatial vector data management: a review

“…The space, time, and spectral resolution are constantly improved, which can be obtained with the TB level, and stored with PB. (2) Surveying and mapping data (Huang et al, 2016;Lu, Yuan, & Yu, 2017;Wang, Guan, & Wu, 2017): it generally includes geographical situation, industry geographic, thematic mapping data, such as 4D (DLG, DRG, DOM, DEM) digital products, land use, and other national basic surveying and mapping data. In recent years, with the 2013Big data giS Big data giS is from offline analysis/ model-driven/ simple visualization to dynamic/ data-driven/ visual analysis in real time generalized giS the characteristics of generalized giS and technical challenges from data collection and cleaning, management and integration, data analysis, and computing Lu and Zhang (2014) Open giS…”