The following information was submitted:
Transactions: SIMULATION AND MODELLING
Transactions ID Number: 10-122
Full Name: Galina Okouneva
Position: Associate Professor
Age: ON
Sex: Female
Address: 350 Victoria Street
Country: CANADA
Tel: 1-416-979-5000
Tel prefix:
Fax:
E-mail address: gokounev@ryerson.ca
Other E-mails:
Title of the Paper: principal component analysis for pose estimation using range data
Authors as they appear in the Paper: G. Oklouneva, D. J. Mctavish, A. Choudhuri
Email addresses of all the authors: gokounev@ryerson.ca, mctavish@ryerson.ca, a2choudh@ryerson.ca
Number of paper pages: 10
Abstract: This paper presents an application of Principal Component Analysis (PCA) to the problem of pose estimation in computer vision. Continuum Shape Constraint Analysis (CSCA), a theoretical development bases on PCA, generates a variety of numerical measures that can be used to assess the shape and predict the accuracy of pose estimation. The proposed approach was developed for LIDAR-based scanning that samples non-specific points from the object across the area observed from a single view. Based on CSCA measures, the paper answers the question: what views of an object can be expected to lead to the lowest pose estimation error computed via the Iterative Closest-Point Algorithm, or conversely, what level of error can be expected for a particular scan view. A novel measure, the Expectivity Index, presented in this paper, is used to assess and predict the pose estimation accuracy. The approach is demonstrated in both numerical simulation and experimental studies using the!
Stanford Bunny, cuboid and asymmetrical cuboctahedron shapes. The continuum nature of the CSCA formulation produces measures that are pure shape properties an object.
Keywords: Computer Vision, LIDAR, ICP, Pose Estimation, Principal Component Analysis
EXTENSION of the file: .doc
Special (Invited) Session: CSCA-Based Expectivity Indices for LIDAR-based Computer Vision
Organizer of the Session: 618-517
How Did you learn about congress: Alexander Pasko, apasko@bournemouth.ac.uk, Anatoly Fomenko, atfomenko@mail.ru, Michael Greenspan, michael.greenspan @queensu.ca
IP ADDRESS: 99.247.8.73
