Previous Next

新闻中心


孙梦溪在 Geoscience Data Journal 上发表文章“A Method for Landslide Deformation Detection Based on Projection Surface Element Matching of 3D Models”

发布时间:2025-04-03 11:50 作者: 来源: 阅读:

孙梦溪在 Geoscience Data Journal 上发表文章“A Method for Landslide Deformation Detection Based on Projection Surface Element Matching of 3D Models


欢迎引用:

Sun, M., Cao, H. and Duan, Y. (2025), A Method for Landslide Deformation Detection Based on Projection Surface Element Matching of 3D Models. Geosci. Data J., 12: e290. https://doi.org/10.1002/gdj3.290




DATA SERVICES ARTICLE
Open Access

A Method for Landslide Deformation Detection Based on Projection Surface Element Matching of 3D Models

First published: 31 March 2025
OpenURL Wuhan University

Funding: This work was supported by National Key Research and Development Program of China, (no. 2023YFB3905704).

ABSTRACT

Landslides represent one of the most prevalent natural disasters worldwide, exerting significant adverse effects on social stability and economic development. Timely and accurate monitoring of landslide changes is crucial for disaster prevention and mitigation. Unlike traditional change detection, which often focus on broad environmental changes, landslide monitoring specifically aims to capture critical parameters such as the precise location of deformation, the direction of movement, and the rate of displacement associated with landslide events. Conventional monitoring techniques are typically constrained to fixed-point observations or are limited to the collection of deformation location data, which may not provide a comprehensive understanding of the landslide's behaviour. To address these limitations, this study proposes an innovative approach for detecting landslide deformation utilising multi-temporal imagery acquired through Unmanned Aerial Vehicles (UAVs). Initially, UAVs are deployed to perform multi-temporal photogrammetric surveys of the landslide-affected area, enabling the construction of high-resolution 3D models. These models facilitate the extraction of the exposed surface by employing advanced vegetation segmentation techniques. Following this, the generated 3D models undergo surface segmentation and normal direction projection, resulting in the creation of orthoimages that accurately represent the slope surface. Subsequently, feature matching is conducted between the orthoimages of the slope surface to identify corresponding points across different temporal datasets. Utilising the forward and inverse transformation relationships of these orthoimages, the deformation direction and velocity of the identified deformation points are calculated. This methodology ultimately enables precise and comprehensive monitoring of landslide deformation. To validate the efficacy of the proposed method, a longitudinal study spanning 4 years was conducted at the Che Yiping landslide site located in western Yunnan Province, China. The findings from this extensive experiment indicate that the proposed approach effectively captures the deformation characteristics of the entire landslide, with point displacement accuracy at specific locations comparable to Global Navigation Satellite System (GNSS) measurements. Furthermore, a detailed analysis of the deformation characteristics within the landslide area revealed significant displacement variations at multiple deformation sites, thereby elucidating the overarching deformation trends present at the landslide location. Through this research, we aim to provide critical data support and a scientific foundation for the prevention of landslide disasters and the management of geological hazards. The insights gained from this study are intended to inform relevant decision-making processes, thereby contributing to enhanced safety and resilience in landslide-prone regions.