基于无人机多光谱的黄土高原植被提取分割分类研究  

Research on Vegetation Extraction,Segmentation and Classification of the Loess Plateau Based on Drone Multi-Spectral

作　　者：李亚北 韩磊[1] LI Yabei;HAN Lei(School of Land Engineering,Chang'an University,Xi'an 710061,China)

机构地区：[1]长安大学土地工程学院,陕西西安710061

出　　处：《河南科技》2024年第4期110-115,共6页Henan Science and Technology

摘　　要：【目的】植被作为生态系统的主要组成部分,其种类和数量及其变化对生态系统有着重要影响。探究在我国黄土高原小流域进行植被提取分割时的最优分割尺度,有助于快速准确地提取植被信息,对于监测黄土高原生态系统状况和维持生态系统稳定具有重要意义。【方法】基于吴起县柴沟流域无人机多光谱影像和面向对象的方法,使用eCognition软件对影像进行多尺度分割研究。【结果】经分析,在分割尺度为240、形状权重为0.7、紧凑权重为0.1时影像的分割效果最好,基于该分割结果,选用纹理特征和光谱特征为分类指标,采用随机森林方法对影像进行分类,分类总体精度和Kappa系数分别为96.2%和0.951。【结论】研究结论可为柴沟流域植被结构优化及黄土高原生态环境保护和植被恢复治理提供技术参考。[Purposes]As the main component of the ecosystem,the type and quantity of vegetation and its changes have an important impact on the ecosystem.Exploring the optimal segmentation scale for vegetation extraction and segmentation in small watersheds of the Loess Plateau in China is helpful to extract vegetation information quickly and accurately,which is of great significance for monitoring the ecosystem status of the Loess Plateau and maintaining ecosystem stability.[Methods]Based on the multi-spectral image and object-oriented method of UAV in Chaigou watershed of Wuqi County,the eCognition software was used to study the multi-scale segmentation of the image.[Findings]After analysis,the segmentation effect of the image is the best when the segmentation scale is 130,the shape weight is 0.5,and the compact weight is 0.5.Based on the segmentation results,the texture features and spectral features are selected as the classification indicators,and the random forest method is used to classify the images.The overall classification accuracy and Kappa coefficient are 96.2%and 0.951,respectively.[Conclusions]The research conclusions can provide technical reference for the optimization of vegetation structure in Chaigou watershed and the ecological environment protection and vegetation restoration in the Loess Plateau.

关 键 词：无人机多光谱 多尺度分割 植被提取 黄土高原小流域 

分 类 号：K903[历史地理—人文地理学]