A 4D Filtering and Calibration Technique for Small-Scale Point Cloud Change Detection with a Terrestrial Laser Scanner

Remote Sensing

Ryan A. Kromer⁴  Antonio Abellán⁴  D. Jean Hutchinson⁴  Matt Lato¹  Tom Edwards⁵  Michel Jaboyedoff³  Marc-Henri Derron²  Richard Müller² 

[1] BGC Engineering, 414 Princeton Ave., Ottawa, ON K2A 1B5, Canada; E-Mail:;Department of Geological Sciences and Geological Engineering, Queen’s University, 36 Union Street, Kingston, ON K7L 3N6, Canada; E-Mail;Risk Analysis Group, Institute of Earth Sciences, University of Lausanne, CH-1015 Lausanne, Switzerland; E-Mails:;Department of Geological Sciences and Geological Engineering, Queen’s University, 36 Union Street, Kingston, ON K7L 3N6, Canada; E-Mail:;Canadian National Railway, 10229–127 Avenue, Edmonton, AB T5E 0B9, Canada; E-Mail:

关键词: point cloud;    de-noising;    LiDAR;    Terrestrial Laser Scanning;    monitoring;    change detection;   

DOI  :  10.3390/rs71013029

来源: mdpi

【 摘 要 】

This study presents a point cloud de-noising and calibration approach that takes advantage of point redundancy in both space and time (4D). The purpose is to detect displacements using terrestrial laser scanner data at the sub-mm scale or smaller, similar to radar systems, for the study of very small natural changes, i.e., pre-failure deformation in rock slopes, small-scale failures or talus flux. The algorithm calculates distances using a multi-scale normal distance approach and uses a set of calibration point clouds to remove systematic errors. The median is used to filter distance values for a neighbourhood in space and time to reduce random type errors. The use of space and time neighbours does need to be optimized to the signal being studied, in order to avoid smoothing in either spatial or temporal domains. This is demonstrated in the application of the algorithm to synthetic and experimental case examples. Optimum combinations of space and time neighbours in practical applications can lead to an improvement of an order or two of magnitude in the level of detection for change, which will greatly improve our ability to detect small changes in many disciplines, such as rock slope pre-failure deformation, deformation in civil infrastructure and small-scale geomorphological change.

【 授权许可】

CC BY   
© 2015 by the authors; licensee MDPI, Basel, Switzerland.