Abstract: Spectral data are the joint affecting results of many factors, and there may be the “same objects with different spectra” or “different objects with the same spectra”. In this study, four common spectral matching algorithms, [WTBX]i.e.[WTBZ] the Minimum Distance (MD), Spectral Similarity Angle (SSA), Spectral Correlation Coefficient (SCC) and Spectral Information Divergence (SID), were used to carry out the matching analysis on the denoised spectral data of plants growing on sand dune in the Horqin Sandy Land. It was found that the minimum distance method could be used to distinguish the typical plant species on sand dune in the study area. On which a matching analysis of spectral derivative was carried out, it was found that the distinguishing effects of spectral similar angle and spectral correlation coefficient were significantly increased, and even a negative value of the matching degree of second derivative spectrum occurred. The band TM1, TM2, TM3 and TM4 of vegetation spectral data were selected to process, and it was verified that the band TM2 and TM4 could be used to improve the distingishing accuracy. After analyzing the basic information of single band with the statistically based analysis, it was found that the amount of information contained in a single band was in an order of TM1>TM4>TM2>TM3, and it was considered that the band TM4 should be the first choice to distinguish the plant species in the study area based on a comprehensive consideration of the highly overlapped spectral reflectance curve and the spectral matching results of single band of plant species.