摘要
滑坡识别是地质灾害防控中的关键环节,有效的滑坡识别方法对灾害防治至关重要,随着遥感技术与人工智能的发展不断革新,滑坡识别的方法和手段也发生了日新月异的变化。系统性概述了滑坡识别的基本理论框架,详细讨论了传统野外调查法、基于像元、面向对象、变化检测、机器学习以及深度学习算法的基本原理。梳理分析了这些方法在滑坡识别中的优势及其存在的局限性。最后,指出了现有研究中存在的主要问题,如高质量数据获取困难、模型泛化能力不足等,并指明了未来滑坡识别方法和技术的研究方向,旨在为滑坡识别方法和技术的研究提供参考。
Effective landslip identification techniques are essential for disaster prevention and control,and landslide identification is a critical component in the prevention and management of geologic hazards.The techniques of landslide identification have changed quickly in tandem with the quick advancement of artificial intelligence and remote sensing technology.This paper systematically summarizes the basic theoretical framework of landslide identification,discusses in detail the fundamentals of the conventional field survey method,image pixel-based,object-oriented,change detection,machine learning,and deep learning algorithms,and analyzes the advantages and limitations of these methods in landslide identification.In conclusion,this paper highlights the key issues present in existing research,such as the challenges in acquiring high-quality data and the insufficient generalization capabilities of models,and suggests the future research direction of landslide identification methods and techniques,aiming to serve as a reference for the field\s studies.
作者
李平
拓耔含
王连升
LI Ping;TUO Zihan;WANG Liansheng(Institute of Disaster Prevention,Sanhe 065201,China;Hebei Provincial Key Laboratory of Earthquake Disaster Prevention and Risk Assessment,Sanhe 065201,China)
出处
《防灾科技学院学报》
2025年第4期82-95,共14页
Journal of Institute of Disaster Prevention
基金
国家重点研发计划(2024YFC3017003)
中央高校基本科研业务费研究生科技创新基金资助(ZY20250341)。
关键词
滑坡识别
遥感技术
人工智能
地质灾害
landslide recognize
remote sensing technique
artificial intelligence
geologic disaster
