基于高光谱遥感的树种识别

Tree Species Recognition Based on Hyperspectral Remote Sensing

高光谱遥感为树种识别提供了新的技术支持,对精准林业发展具有重要意义。该研究以USGS光谱库树种冠层光谱为数据源,分析不同数学变换形式和常用的植被指数对树种识别能力的影响。结果表明:5种树种原始光谱反射率在350~500 nm的波长范围内存在明显差别,橄榄树反射率最高,750~1 400 nm波长原始反射光谱曲线差异最为明显;采用不同数学变换形式识别该5种树种的最佳波长位置,一阶微分为411 nm、949 nm、1 143 nm、1 393 nm、1 885 nm、2 315 nm和1 508 nm,二阶微分为533 nm、694 nm、742 nm、1 133 nm、1 383 nm、1 408 nm、1 865 nm和1 895 nm,去除包络线后为490 nm,1 453 nm和1 915 nm等;采用树种冠层原始光谱计算的几种植被指数中CRI辨识树种的能力最强。由此可见,利用不同特征波段和植被指数能够进行树种识别。

Hyperspectral remote sensing provides a new technical support for tree species identification,and it is of great significance for the development of precision forestry. The tree species canopy spectrum of the USGS spectral library was used as the data source to analyze the ability of different mathematical transformations and commonly used vegetation indices to identify tree species. The results show that there are obvious differences in the original spectral reflectance of the five tree species in the wavelength range of 350~500 nm. The olive tree has the highest reflectance, and the difference of the original reflectance spectrumcurve of the wavelength of 750~1 400 nm is the most obvious. These five species are identified by different mathematical transformation, the optimal wavelength position of the tree species is 411 nm, 949 nm, 1 143 nm, 1 393 nm, 1 885 nm, 2 315 nm, and 1 508 nm after first order differential transformation, and 533 nm, 694 nm, 742 nm, 1 133 nm, 1 383 nm, 1 408 nm, 1 865 nm and 1 895 nm after second order differential transformation, after removing the envelope curve, 490 nm, 1453 nm and 1915 nm, etc. The CRI has the strongest ability to identify tree species among several vegetation indices calculated from the original spectrum of the tree canopy. It can be seen that different characteristic bands and vegetation indices can identify tree species.