Skip to Main content Skip to Navigation
Theses

Terrestrial Laser Scanner Noise Analysis, Modelling and Detection

Abstract : In this thesis, we focused on several topics related to noise detection in point cloud generated by Terrestrial Laser Scanners (TLS). First, the projection methods to compute an image from a TLS scan. Second, the detection of sky noise, i.e. noise produced when a Amplitude Modulated Continuous Wave TLS measures range only from background radiation. And finally, the detection of mixed point noise, i.e. points acquired when the TLS was receiving return signals from several different surfaces. To tackle these challenges, we first analysed how the TLS samples space and deduced properties on how the local point cloud density evolves with respect to the elevation, this allowed us to show the limits of usual noise detection techniques and oriented our focus on 2D non density based detection techniques. We then defined a theoretical framework to analyse projection methods, unavoidable foundations for 2D detection methods. This framework allowed us to bring to light two fundamental properties that should be satisfied by a projection. Following these properties, we designed a projection algorithm that satisfied them as much as possible. We then defined a way to quantify projection quality and compared our proposed algorithm with the widely used classic algorithm and showed that the classic projection method is not adapted. Our proposed projection however showed very good results. Since the sky noise was never studied in previous works, we formally analysed it to build some theoretical foundations for sky detection. The analysis allowed us to show theoretically and experimentally that the range distribution of sky noise is independent of the underlying properties of the background radiation signal. From our projection and the discovered properties, we designed a sky detector and a mixed point detector. The detectors were tested via an extensive validation in controlled conditions. The results showed that our proposed detectors combined with the proposed projection are able to correctly detect almost all presented noise with few bad detection for the sky detectors and reasonable amount for the mixed point detector.
Document type :
Theses
Complete list of metadata

Cited literature [49 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-02626899
Contributor : Abes Star :  Contact
Submitted on : Tuesday, May 26, 2020 - 6:21:08 PM
Last modification on : Wednesday, November 3, 2021 - 6:44:58 AM

File

ROMBOURG_2019_archivage.pdf
Version validated by the jury (STAR)

Identifiers

  • HAL Id : tel-02626899, version 1

Collections

Citation

Romain Rombourg. Terrestrial Laser Scanner Noise Analysis, Modelling and Detection. Signal and Image Processing. Université Grenoble Alpes, 2019. English. ⟨NNT : 2019GREAM064⟩. ⟨tel-02626899⟩

Share

Metrics

Record views

308

Files downloads

156