A 4D photogrammetric system for studying flow-like landslides

This thesis presents a photogrammetric system developed for experimental studies of laboratory landslide models. The hardware captures granular flows as time-series three-dimensional (4D) point clouds, producing high-quality digital datasets of laboratory landslide experiments. The software incorporates a robust tracking algorithm, enabling the calculation of flow velocity vectors at the single-particle or grain scale with high spatial and temporal resolution, even amid the fast and tumbling flows. We begin by describing the development of the photogrammetric hardware, which forms the foundation of this research. The core component is a stereo camera system consisting of two setups: a high-speed system optimized for motion tracking and a standard-speed system optimized for static scene scanning. Additionally, we discuss the peripheral hardware essential for high-speed imaging and laboratory landslide experiments. Next, we address the control software, which ensures synchronization across multiple cameras—a crucial aspect of a stereo vision system. We also outline the recommended system parameters, providing guidelines for configuring high-speed camera systems in a laboratory environment. The core software breakthrough lies in the implementation of a tracking algorithm for single-particle motion in granular flows. This algorithm utilizes several computer vision techniques to preprocess raw image data. Building on this foundation, we outline the steps of the original Shade-SIFT Matching algorithm and the Chain Tracking algorithm, which achieves particle tracking in two-dimensional images and three-dimensional space, respectively. The target application of this research is measuring movement distribution in landslide experiments. In the concluding stage of the study, the system’s capability for surface topology analysis and velocity mapping is demonstrated through granular column collapse flume experiments.

Name dc:type.qualificationname

Doctor of Philosophy (PhD)

Level dc:type.qualificationlevel

Doctoral

Grantor dc:publisher.institution

University of Cambridge

Year dc:date.issued

2025

Author dc:creator

• He, Zhiwei

Advisor dc:contributor.advisor

• Liang, Dongfang

• 4D
• Photogrammetry
• Flow-like Landslide

Licence

• https://www.repository.cam.ac.uk/bitstreams/5cbba5ec-f289-4b35-a78a-02213c72c843/download
• http://purl.org/NET/rdflicense/allrightsreserved

Language dc:language

eng

DOI dc:identifier.doi

https://doi.org/10.17863/CAM.121232

OAI identifier oai:identifier

oai:www.repository.cam.ac.uk:1810/389254