DocumentCode :
1662670
Title :
A 3D classifier trained without field samples
Author :
Douillard, Bertrand ; Quadros, A. ; Morton, Peter ; Underwood, J.P. ; De Deuge, Mark
Author_Institution :
Australian Centre for Field Robot., Univ. of Sydney, Sydney, NSW, Australia
fYear :
2012
Firstpage :
805
Lastpage :
810
Abstract :
This paper presents a 3D classifier that is shown to maintain performance whether trained with real sensor data from the field or purely trained with 3D geometric (Computer Aided Design, CAD, like) models (downloaded from the Internet for instance). The proposed classifier is a global 3D template matching technique which exploits the location of the ground surface for more accurate alignment. The segmentation and position of the ground is given by the segmentation technique in [7] (which does not assumed the ground to be flat). The proposed classifier outperforms Spin Image and Fast Point Feature Histogram (FPFH) based classifiers by up to 30% (the latter being tested at different scales), in the case of sparse 3D data acquired with a Velodyne sensor. In addition, the experimental results suggest that field samples may not be required in the training set of alignment-based 3D classifiers. This finding implies that the laborious task of gathering hand labelled field data for training may be avoidable for this type of classifier.
Keywords :
CAD; computational geometry; image classification; image segmentation; solid modelling; 3D geometric model; CAD; Velodyne sensor; alignment-based 3D classifier; computer aided design; global 3D template matching; segmentation technique; sparse 3D data; Accuracy; Buildings; Iterative closest point algorithm; Principal component analysis; Shape; Solid modeling; Training;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Control Automation Robotics & Vision (ICARCV), 2012 12th International Conference on
Conference_Location :
Guangzhou
Print_ISBN :
978-1-4673-1871-6
Electronic_ISBN :
978-1-4673-1870-9
Type :
conf
DOI :
10.1109/ICARCV.2012.6485261
Filename :
6485261
Link To Document :
بازگشت