NCSI: An Efficient Parallel Method for Large-Scale 3D Point Cloud Registration

Status	Completed
Mentor Name	Hao Ji
Mentor's XSEDE Affiliation	Education Allocation
Mentor Has Been in XSEDE Community	4-5 years
Project Title	An Efficient Parallel Method for Large-Scale 3D Point Cloud Registration
Summary	Point cloud registration plays an important role in many computer vision applications such as 3D object reconstruction and object detection/segmentation. However, when dealing with a large number of point clouds, many existing registration algorithms appear to have either high computational cost or limited scalability to modern parallel and distributed computing systems. In this project, we will investigate an accelerated approach to rapidly aligning multiple point clouds to a globally consistent structure by taking advantage of parallel computing. We will benchmark the proposed registration approach using real-life applications such as 3D object reconstruction with depth sensors.
Job Description	The student will study efficient registration approaches and develop an accelerated implementation for point cloud registration using parallel computing. The student will also conduct experiments to benchmark the proposed registration approach using 3D object reconstruction applications.
Computational Resources	XSEDE Bridges GPU Cluster and Bridges Regular Memory Cluster.
Contribution to Community	The project targets a parallel approach to rapidly aligning large-scale point cloud data using the XSEDE computing clusters. Students participating in this project will gain practical skills in processing 3D point cloud data with parallel computing. More important, we expect that this project could help them realize the importance and potential of high-performance computing in processing large-scale computations and stimulate their interest in developing modern scalable algorithms on XSEDE clusters for data intelligence applications.
Position Type	Intern
Training Plan	I will teach the student about the algorithms for 3D point cloud registration and how to use XSEDE clusters. During this project, I will have weekly meetings with the student to guide research progress.
Student Prerequisites/Conditions/Qualifications
Duration	Semester
Start Date	01/23/2021
End Date	05/14/2021