To read this content please select one of the options below:

New methodologies for precise building boundary extraction from LiDAR data and high resolution image

Hui Li (Planetary Research Institute, China University of Geosciences, Wuhan, China)
Cheng Zhong (Three Gorges Research Center for Geo‐hazard, Ministry of Education China University of Geosciences, Wuhan, China)
Xiaoguang Hu (Peking University, Beijing, China)
Long Xiao (Planetary Research Institute, China University of Geosciences, Wuhan, China)
Xianfeng Huang (State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan, China)

Sensor Review

ISSN: 0260-2288

Article publication date: 22 March 2013

639

Abstract

Purpose

Light Detection and Ranging (LiDAR) offers a fast and effective way to acquire DSM and extract ground objects such as building, trees and so on. However, it is difficult to extract sharp and precise building boundary from LiDAR data, because its ground sample distance (GSD) is often worse than that of high resolution image. Recently, fusion of LiDAR and high resolution image becomes a promising approach to extract precise boundary. To find the correct and precise boundary, the aim of this paper is to present a series of novel algorithms to improve the quality.

Design/methodology/approach

To find the correct and precise boundary, this paper presents a series of novel algorithms to improve the quality. At first, a progressive algorithm is presented to register LiDAR data and images; second, a modified adaptive TIN algorithm is presented to filter ground point, where a region growth method is applied in the adaptive TIN algorithm; third, a novel criterion based on the density, connectivity and distribution of point cluster is developed to distinguish trees point; fourth, a novel method based on the height difference between neighbor points is employed to extract coarse boundaries; at last, a knowledge based rule is put forward to identify correct building boundary from parallel edges.

Findings

Thorough experiments, it is conducted that: the registration results are accurate and reliable; filtered ground points has good quality, without missing or redundancy; all tree clusters bigger than one grid are detected, and points of walls and edges are eliminated with the new criterion; detected edges exactly locate at real building boundaries, and statistics show the detection correctness is 98 percent, and the detection completeness is 95 percent.

Originality/value

All results prove that precise boundary can be extracted with fusion of LiDAR and high resolution image.

Keywords

Citation

Li, H., Zhong, C., Hu, X., Xiao, L. and Huang, X. (2013), "New methodologies for precise building boundary extraction from LiDAR data and high resolution image", Sensor Review, Vol. 33 No. 2, pp. 157-165. https://doi.org/10.1108/02602281311299699

Publisher

:

Emerald Group Publishing Limited

Copyright © 2013, Emerald Group Publishing Limited

Related articles