Vol.:(0123456789)
The Journal of Supercomputing
https://doi.org/10.1007/s11227-019-02830-9
1 3
3D object recognition method with multiple feature
extraction from LiDAR point clouds
Yifei Tian
1,2
· Wei Song
1,3
· Su Sun
1
· Simon Fong
2
· Shuanghui Zou
1
© Springer Science+Business Media, LLC, part of Springer Nature 2019
Abstract
During autonomous driving, fast and accurate object recognition supports envi-
ronment perception for local path planning of unmanned ground vehicles. Feature
extraction and object recognition from large-scale 3D point clouds incur massive
computational and time costs. To implement fast environment perception, this paper
proposes a 3D recognition system with multiple feature extraction from light detec-
tion and ranging point clouds modifed by parallel computing. Efective object
feature extraction is a necessary step prior to executing an object recognition pro-
cedure. In the proposed system, multiple geometry features of a point cloud that
resides in corresponding voxels are computed concurrently. In addition, a scale flter
is employed to convert feature vectors from uncertain count voxels to a normalized
object feature matrix, which is convenient for object-recognizing classifers. After
generating the object feature matrices of all voxels, an initialized multilayer neural
network (NN) model is trained ofine through a large number of iterations. Using
the trained NN model, real-time object recognition is realized using parallel com-
puting technology to accelerate computation.
Keywords 3D object recognition · Feature extraction · LiDAR point cloud · Parallel
computing
* Wei Song
sw@ncut.edu.cn
1
North China University of Technology, No. 5 Jinyuanzhuang Road, Shijingshan District,
Beijing 100-144, China
2
Department of Computer and Information Science, University of Macau, Taipa 999-078,
Macau, China
3
Beijing Key Lab on Urban Intelligent Trafc Control Technology, Beijing 100-144, China