基于eCognition影像解译技术的风景环境 植被景观遥感量化研究——以云南鸡足山为例  

An RS Quantitative Research of Vegetation Landscape in Scenic Environment Based on eCognition Image Interpretation Technology:A Case Study of Jizu Mountain,Yunnan Province

作　　者：张学玲 ZHANG Xueling

机构地区：[1]北京建筑大学建筑与城市规划学院北京未来城市设计高精尖创新中心,北京100044

出　　处：《中国园林》2020年第10期81-85,共5页Chinese Landscape Architecture

基　　金：中国博士后科学基金面上项目(编号2018M641363)资助。

摘　　要：风景环境多以植被为景观基质,构成具有特定绿化格局和景观特征的环境综合体,相关研究正由定性发展至定量,由图形识别发展至数字分析与算法描述。利用eCognition高分遥感技术进行植被景观影像识别与特征辨析,正迅速以卫星视野提供一种更科学、更宏观的研究途径与分析方法。以鸡足山核心风景区植被景观为例,以中国资源3号卫星遥感影像数据为基础,采用eCognition面向对象分类方法进行量化解析研究,选用光谱异质性、多尺度分割函数和归一化植被指数作为图像解析标准,通过隶属度函数推导获取适用于风景园林研究的数学算法,最终有效提取出针叶林等6类植被景观特征要素信息,科学呈现鸡足山植被景观分布情况,实现了植被特征的算法描述与图纸生成,分类精度经误差矩阵评价结果良好。相关成果为当代景观高空间解析、城市绿地系统分析提供了一种量化调查与分析方法,也为相关人居环境科学研究提供了可参照的景观遥感解译平台与专项技术。Mostly based on the green scenery,the landscape environment is characterized by an environmental complex of specific greening patterns and landscape features.Relevant research on the vegetation landscape environment is being transformed from qualitative analysis to quantitative calculation,with pattern recognition being developed into digital analysis and algorithm description.The recognition and feature analysis of vegetation landscape using eCognition RS technology provides more scientific and macroscopic research approaches and analysis techniques.With the vegetation landscape of the scenic area of Jizu Mountain as an example,the eCognition object-oriented classification method is used for quantitative analysis and research based on the multi-spectral remote sensing image data of China Resources No.3 Satellite.With the spectral heterogeneity,multi-resolution segmentation function and Normalized Difference Vegetation Index(NDVI)as standards of image analysis,the mathematical algorithm suitable for the research of landscape garden is obtained by the Membership Functions,and finally the features of six kinds of vegetation landscapes such as the coniferous forest are extracted.The distribution of the vegetation landscape in Jizu Mountain is scientifically presented and the algorithm description and drawing generation of vegetation characteristics are realized.Evaluated by the error matrix,the classification accuracy is great.The study provides a quantitative survey and analysis method for high spatial analysis of contemporary landscapes and analysis;and also provides valuable and promising remote sensing interpretation platforms and special technologies for scientific research on human settlements.

关 键 词：风景园林 植被景观 ECOGNITION 影像解译技术 遥感 鸡足山 

分 类 号：TU986[建筑科学—城市规划与设计]