深度学习在植被高光谱遥感中的研究综述  

A Review of Deep Learning in Hyperspectral Remote Sensing of Vegetation

作　　者：王俊迪 赵慧[2] 罗耀华[1] 袁正蓉 WANG Jundi;ZHAO Hui;LUO Yaohua;YUAN Zhengrong(College of Computer Science and Cyber Security,Chengdu University of Technology,Chengdu 610059,China;Institute of Mountain Hazards and Environment,CAS,Chengdu 610299,China;University of Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]成都理工大学计算机与网络安全学院,成都610059 [2]中国科学院、水利部成都山地灾害与环境研究所,成都610299 [3]中国科学院大学,北京100049

出　　处：《遥感信息》2024年第4期11-22,共12页Remote Sensing Information

基　　金：科技部重大专项(2019QZKK0404);西藏科技厅重大专项(XZ202201ZD0005G02)。

摘　　要：高光谱遥感影像的光谱分辨率高,在植被生态监测领域具有极大的应用潜力,不仅能够准确区分植被类型,还能够对植被参数进行精准反演。海量高维的高光谱数据应用于植被监测需要高效、准确、灵活的数据分析和计算方法,深度学习为高光谱遥感在植被研究中存在的维度灾难、“同谱异物”和“同物异谱”、非线性特征提取等问题提供了一种新的解决方法。文章首先从计算机算法结构上阐明了深度学习模型在高光谱遥感应用中的优缺点,其次从植被分类和参数反演两方面综述了深度学习算法在高光谱的应用场景,最后指出了深度学习应用存在的问题并提出未来的研究趋势。Hyperspectral remote sensing images have high spectral resolution and great potential for application in the field of vegetation ecological monitoring.They can not only accurately distinguish vegetation types,but also accurately invert vegetation parameters.The application of massive high-dimensional hyperspectral data for vegetation monitoring requires efficient,accurate,flexible data analysis and calculation methods.Deep learning provides a new solution to the problems of dimensionality curse,“spectral variability among different objects”and“spectral variability within the same object”,as well as nonlinear feature extraction in vegetation research using hyperspectral remote sensing.Firstly,this article elucidates the advantages and disadvantages of deep learning models in hyperspectral remote sensing applications from the perspective of computer algorithm structure.Secondly,the application scenarios of deep learning algorithms in hyperspectral analysis are summarized from two aspects,which are vegetation classification and parameter inversion.Finally,it point outs the problems existing in deep learning applications and proposes future research trends.

关 键 词：深度学习 植被高光谱 物种识别 理化参数 遥感反演 

分 类 号：TP79[自动化与计算机技术—检测技术与自动化装置]