Analysis of Hyper-spectral Discriminatory Methods on Typical Plateau Forestry Vegetation

Spectral characteristics analyzing is the foundation of classification and matching for surface features. In this paper, field spectrum measurement was conducted on typical plateau forestry vegetation types in upper reaches of Min Jiang River. Meanwhile, hyper-spectral similarity measurement indicators were established. This paper tested 5 spectral similarity measurement operators including Euclidean Distance (ED), Spectral Angel Measurement (SAM), Spectral Information Divergence (SID), Spectral Information Divergence-Spectral Angel Measurement (SID-SAM) and Douglas Peucker Spectral Dimensionality Reduction Distance (DPSD) on identification of plateau forestry vegetation in quantified analyzing methods. Analyzing results indicate that (1) the spectral characteristics differences of all the 5 plateau forestry vegetation types mainly fall into the crest or trough of reflection spectra; (2) Chinese China fir, palm, pilea notata, ferns and mottled bamboo generally have high standard of spectral discriminatory probability, meanwhile, the spectral discriminatory probability between pilea notata and ferns is quite low; (3)for plateau forestry vegetation types, spectral discriminatory ability for of all the 5 spectral similarity measurement operators shows in descending order: Spectral Angel Measurement, Euclidean Distance Measurement, Douglas Peucker Spectral Dimensionality Reduction Distance Measurement, Spectral Information Divergence- Spectral Angel Measurement, spectral Information Divergence Measurement.