On creating low dimensional 3D feature descriptors with PCA

With the availability of commodity depth sensors and depth sensing capabilities on mobile devices, there is a need to develop memory efficient and computationally cheap applications. 3D feature descriptor based keypoint matching is the very first step in various computer vision and robotic applicati...

Full description

Saved in:
Bibliographic Details
Main Authors: Prakhya, Sai Manoj, Liu, Bingbing, Lin, Weisi, Li, Kun, Xiao, Yong
Other Authors: School of Computer Science and Engineering
Format: Conference or Workshop Item
Language:English
Published: 2020
Subjects:
Online Access:https://hdl.handle.net/10356/140477
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
id sg-ntu-dr.10356-140477
record_format dspace
spelling sg-ntu-dr.10356-1404772020-05-29T07:36:44Z On creating low dimensional 3D feature descriptors with PCA Prakhya, Sai Manoj Liu, Bingbing Lin, Weisi Li, Kun Xiao, Yong School of Computer Science and Engineering 2017 IEEE Region 10 Conference (TENCON 2017) Engineering::Computer science and engineering Three-dimensional Displays Principal Component Analysis With the availability of commodity depth sensors and depth sensing capabilities on mobile devices, there is a need to develop memory efficient and computationally cheap applications. 3D feature descriptor based keypoint matching is the very first step in various computer vision and robotic applications, and is also one of the most memory and computationally demanding steps. In this work, we apply Principal Component Analysis (PCA) to reduce the dimensionality of 3D feature descriptors and show that they retain their descriptiveness, while drastically decreasing their memory and computational requirements. We apply this scalable PCA based dimensionality reduction technique on three state-of-the-art 3D feature descriptors and quantitatively evaluate their performance with varying dimensionality. Experimental results on publicly available Bologna Kinect dataset show that the proposed method reduces the dimensionality of SHOT, RoPS and FPFH feature descriptors from 352 to 50, 135 to 30 and 33 to 15, respectively, while offering more than 90% of the keypoint matching performance. ASTAR (Agency for Sci., Tech. and Research, S’pore) 2020-05-29T07:36:43Z 2020-05-29T07:36:43Z 2017 Conference Paper Prakhya, S. M., Liu, B., Lin, W., Li, K., & Xiao, Y. (2017). On creating low dimensional 3D feature descriptors with PCA. Proceedings of 2017 IEEE Region 10 Conference (TENCON 2017), 315-320. doi:10.1109/TENCON.2017.8227882 978-1-5090-1135-3 https://hdl.handle.net/10356/140477 10.1109/TENCON.2017.8227882 2-s2.0-85044219643 315 320 en © 2017 IEEE. All rights reserved.
institution Nanyang Technological University
building NTU Library
country Singapore
collection DR-NTU
language English
topic Engineering::Computer science and engineering
Three-dimensional Displays
Principal Component Analysis
spellingShingle Engineering::Computer science and engineering
Three-dimensional Displays
Principal Component Analysis
Prakhya, Sai Manoj
Liu, Bingbing
Lin, Weisi
Li, Kun
Xiao, Yong
On creating low dimensional 3D feature descriptors with PCA
description With the availability of commodity depth sensors and depth sensing capabilities on mobile devices, there is a need to develop memory efficient and computationally cheap applications. 3D feature descriptor based keypoint matching is the very first step in various computer vision and robotic applications, and is also one of the most memory and computationally demanding steps. In this work, we apply Principal Component Analysis (PCA) to reduce the dimensionality of 3D feature descriptors and show that they retain their descriptiveness, while drastically decreasing their memory and computational requirements. We apply this scalable PCA based dimensionality reduction technique on three state-of-the-art 3D feature descriptors and quantitatively evaluate their performance with varying dimensionality. Experimental results on publicly available Bologna Kinect dataset show that the proposed method reduces the dimensionality of SHOT, RoPS and FPFH feature descriptors from 352 to 50, 135 to 30 and 33 to 15, respectively, while offering more than 90% of the keypoint matching performance.
author2 School of Computer Science and Engineering
author_facet School of Computer Science and Engineering
Prakhya, Sai Manoj
Liu, Bingbing
Lin, Weisi
Li, Kun
Xiao, Yong
format Conference or Workshop Item
author Prakhya, Sai Manoj
Liu, Bingbing
Lin, Weisi
Li, Kun
Xiao, Yong
author_sort Prakhya, Sai Manoj
title On creating low dimensional 3D feature descriptors with PCA
title_short On creating low dimensional 3D feature descriptors with PCA
title_full On creating low dimensional 3D feature descriptors with PCA
title_fullStr On creating low dimensional 3D feature descriptors with PCA
title_full_unstemmed On creating low dimensional 3D feature descriptors with PCA
title_sort on creating low dimensional 3d feature descriptors with pca
publishDate 2020
url https://hdl.handle.net/10356/140477
_version_ 1681059198365859840