Citation: Wang, X.; Lyu, H.; He, W.;
Chen, Q. Line Structure Extraction
from LiDAR Point Cloud Based on
the Persistence of Tensor Feature.
Appl. Sci. 2022, 12, 9190. https://
doi.org/10.3390/app12189190
Academic Editors: M. Jamal Deen,
Subhas Mukhopadhyay,
Yangquan Chen, Simone Morais,
Nunzio Cennamo and Junseop Lee
Received: 1 August 2022
Accepted: 31 August 2022
Published: 14 September 2022
Publisher’s Note: MDPI stays neutral
with regard to jurisdictional claims in
published maps and institutional affil-
iations.
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Article
Line Structure Extraction from LiDAR Point Cloud Based on the
Persistence of Tensor Feature
Xuan Wang
1
, Haiyang Lyu
2
, Weiji He
1,
* and Qian Chen
1
1
Jiangsu Key Laboratory of Spectral Imaging & Intelligence Sense (SIIS), Nanjing University of Science
and Technology, Nanjing 210094, China
2
Smart Health Big Data Analysis and Location Services Engineering Research Center of Jiangsu Province,
Nanjing University of Posts and Telecommunications, Nanjing 210023, China
* Correspondence: hewj@njust.edu.cn; Tel.: +86-025-85866638
Featured Application: line structure extraction, tensor voting, persistent homology.
Abstract:
The LiDAR point cloud has been widely used in scenarios of automatic driving, object
recognition, structure reconstruction, etc., while it remains a challenging problem in line structure
extraction, due to the noise and accuracy, especially in data acquired by consumer electronic devices.
To address the issue, a line structure extraction method based on the persistence of tensor feature is
proposed, and subsequently applied to the data acquired by an iPhone-based LiDAR sensor. The
tensor of each point is encoded, voted, and aggregated by its neighborhood, and further decomposed
into different geometric features in each dimension. Then, the line feature in the point cloud is repre-
sented and computed using the persistence of the tensor feature. Finally, the line structure is extracted
based on the persistent homology according to the discrete Morse theory. With the LiDAR point
cloud collected by the iPhone 12 Pro MAX, experiments are conducted, line structures are extracted
from two different datasets, and results perform well in comparison with other related results.
Keywords:
line structure extraction; tensor feature decomposition; persistent homology; iPhone-
based LiDAR sensor
1. Introduction
Light detection and ranging (LiDAR) is a method to measure the distance between the
object and the receiver based on the reflection of light and obtain massive points for the sur-
face of an area instantly [
1
–
3
]. With high efficiency and versatile performance, it has been ex-
tensively applied in surveying and mapping, automatic driving, scene
understanding [4,5]
,
etc. In recent years, LiDAR sensors have been equipped in consumer electronic devices,
such as Kinect, iPhone, etc., which has made it more and more convenient to acquire point
cloud data for common users, and many applications have been constructed [
6
,
7
], such as
line structure extraction, shape recognition, object detection and classification, and some
high-level applications (scene understanding, simultaneous localization and mapping, etc.).
With these easy-to-use LiDAR devices, plenty of point cloud datasets are provided,
and abundant information can be extracted, such as geometric feature, semantic labeling,
and scenario relations [
8
]. However, it also brings many problems in dealing with these re-
dundant point cloud datasets, since not all points are needed, and it’s difficult to extract the
structure information, compress the data, and represent these redundant datasets by simple
geometric structures [
9
,
10
]. To solve these problems and obtain structure information,
different studies have been conducted using, e.g., deep learning-based feature extraction
methods and geometric model fitting methods [
7
,
8
,
11
,
12
]. These methods either need prede-
fined geometric models [
12
,
13
] or large amounts of training datasets [
7
]. In addition, some
post-processing operations are also needed to maintain connection relations of geometric
structures, and results can be affected by the quality of the point cloud datasets [
14
,
15
].
Appl. Sci. 2022, 12, 9190. https://doi.org/10.3390/app12189190 https://www.mdpi.com/journal/applsci
