计算机应用与软件2016，Vol.33Issue(11)：185-187,192,4.DOI:10.3969/j.issn.1000-386x.2016.11.043
基于OpenCL的ICP点云并行配准算法
PARALLEL REGISTRATION ALGORITHM OF ICP POINT CLOUD BASED ON OPENCL
摘要
Abstract
In view of the problem of too low efficiency the current point cloud registration algorithm has,by using OpenCL we achieved the general GPU-based kd-tree parallel search algorithm,and then realised the parallel ICP point cloud registration algorithm.First we established the three-dimensional kd-tree of target point cloud,and used the OpenCL to parallelly accelerate its search algorithm.Then we applied the parallelly accelerated kd-tree search algorithm in ICP algorithm,at the same time the OpenCL parallel acceleration was also used aimed at the rest part of ICP algorithm to ensure as much as possible the efficient registration process.Experiments verified the efficiency and robustness of the implemented algorithm.关键词
OpenCL/GPU/kd-tree/点云配准/ICPKey words
OpenCL/GPU/kd-tree/Point cloud registration/ICP分类
信息技术与安全科学引用本文复制引用
刘忠建..基于OpenCL的ICP点云并行配准算法[J].计算机应用与软件,2016,33(11):185-187,192,4.基金项目
国家自然科学基金项目（61202229）；北京市自然科学基金项目（4132018）。 （）
