霍江润 李晶 曹泽远 夏颖聪 李珂 王子涵

摘  要：植被识别和特征研究是智慧校园的有机组成部分。该文应用高光谱遥感设备对中国矿业大学（北京）学院路校区20种典型植被的反射光谱数据进行采集并对反射光谱基本特征、光谱时序变化、相关植被指数以及低通滤波变换进行分析，研究结论：（1）植被反射光谱基本特征相似，紫叶李在可见光波段的波峰位置存在偏移，八宝和黄杨的反射率波形特征明显;（2）植被反射率时序变化与物候规律较为一致，地毯草和果树类存在偏差;（3）银杏、紫叶李、刚竹、木槿、忍冬、石榴和八宝的NDVI特征明显，悬铃树、槭树、刚竹、柏树、叉子圆柏、山桃、八宝以及爬山虎的EVI特征明显;（4）柏树、紫藤和悬铃树的低通滤波反射特征明显。研究成果有助于植被的辨识，为智慧校园在植被管理建设方面奠定了基础。

关键词：高光谱 校园植被 树种识别 光谱特征变换

中图分类号：P237  文献标识码：A   文章编号：1672-3791（2021）10（b）-0000-00

Campus Vegetation Recognition Based on Measured Hyper-spectral Data

—A Case Study from China University of Mining & Technology， Beijing

HUO Jiangrun1，2  LI Jing1* CAO Zeyuan1  XIA Yingcong1  LI Ke1  WANG Zihan1

（College of Geosciences and Surveying Engineering， China University of Mining and Technology （Beijing）， Beijing， 100083 China）

Abstract： Vegetation recognition and characteristic research is an integral part of smart campus. In this paper， hyperspectral remote sensing equipment is used to collect the reflectance spectral data of 20 typical vegetation on The College Road campus of China University of Mining and Technology （Beijing）， and it analyzes the basic characteristics of reflectance spectrum， spectral time series changes， relevant vegetation indices and low-pass filtering transformation. The research conclusions are as follows： （1）The basic characteristics of reflectance spectra of vegetation are similar， the peak position of purple plum in visible band is offset， and the reflectance waveform characteristics of Baopai and Boxwood are obvious; （2）The temporal variation of vegetation reflectance was consistent with phenology， but there were deviations between carpet grass and fruit trees. （3）The NDVI characteristics of Gin kgo biloba， Prunus purpura， Phyllostachys japonicae， Hibiscus japonicae， Pomegranate and Eight species were obvious， and the EVI characteristics of camphora， Maple， Phyllostachys japonicae， cypress， Cypress， Cypress， Mountain peach， Eight species and Ivy were obvious. （4）Cypress， Wisteria and camellia have obvious low-pass filter reflection characteristics. The research results contribute to the identification of vegetation and lay a foundation for the construction of vegetation management in smart campus.

Key Words： Hyper-spectral; Campus vegetation; Tree species identification; Spectral feature transformation

高光譜遥感（Hyperspectral Remote Sensing）技术将表征地物属性特征的光谱信息与表征地物几何位置关系的空间信息有机地结合起来，使得地物的精准分析与细节提取成为了可能[1]。

在高光谱遥感发展的过程中，国内外在植被方面进行了大量研究。国外建立的植被光谱数据库主要有：1960～1970年，美国NASA建立地球资源信息光谱数据库[2];约翰斯·霍普金斯大学建立的光谱数据库;美国地质调查局建立的地物光谱数据库;美国环保局部门建立的AEDC/EPA光谱数据库及森林高光谱数据库等。……