A Review of Deep Learning in Hyperspectral Remote Sensing of Vegetation
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.
deep learninghyperspectral of vegetationspecies identificationphysicochemical parameterremote sensing inversion
万方数据
维普
