基于遥感影像的土地利用变化检测方法  

Land use change detection method based on remote sensing images

作　　者：肖芬 XIAO Fen(The Natural Resources Bureau of Kaili City,Kaili,Guizhou 556000,China)

机构地区：[1]凯里市自然资源局,贵州凯里556000

出　　处：《北京测绘》2025年第3期345-349,共5页Beijing Surveying and Mapping

摘　　要：针对土地利用变化检测存在的现势性与准确性低的问题,本文采用一种针对土地利用现状(CLUS)的年度变化检测方法,以2020年WorldView-2(WV2)和2021年SuperView-1(SV1)两期高分辨率遥感影像为数据源,采用经优化的双峰分裂阈值法识别和消除建筑物阴影的干扰,为接下来的变化检测工作提供坚实的基础。然后基于孪生神经网络(SNN)构建一个变化检测模型,并基于地理信息系统(GIS)优化处理,从而实现对土地利用现状变化区域的迅速定位。实验结果表明,本文提出的方法能够快速且准确地识别出不同时期影像中土地利用现状的变化位置,在保持轻量化的同时表现出突出的检测性能,为相关类变化检测的研究与应用提供参考。To address the issues of low currency and accuracy in land use change detection,this paper employed an annual change detection approach tailored to the current land use status(CLUS).By utilizing two sets of high-resolution remote sensing images from WorldView-2(WV2)in 2020 and SuperView-1(SV1)in 2021 as data sources,the paper applied an optimized bimodal split threshold method to identify and eliminate the interference of building shadows,thereby laying a solid foundation for subsequent change detection tasks.Subsequently,by leveraging a siamese neural network(SNN),a change detection model was constructed and optimized through geographic information system(GIS)processing,so as to rapidly locate regions with varying CLUS.Experimental results demonstrate that the proposed method can rapidly and accurately identify and locate the regions with varying CLUS in images collected at different periods,exhibiting remarkable detection performance while maintaining a lightweight nature.This paper provides a reference for research and applications related to similar change detection tasks.

关 键 词：影像变化检测(ICD) 阴影检测(SD) 土地利用现状(CLUS) 孪生神经网络(SNN) 

分 类 号：P237[天文地球—摄影测量与遥感]