地物光谱匹配模型比较研究

COMPARISON OF DIFFERENT SPECTRAL MATCH MODELS

在大量实测的城市地物光谱数据的基础上,对3种有代表性的光谱匹配模型,即最小距离、光谱角度匹配(SAM),光谱相似度匹配进行了深入计算和分析,在这3种模型中,以SAM表现程度较好,但都不是最理想,区分程度不是很高.为增强区别程度,本文引入了导数对光谱匹配模型进行了重新计算,分析结果表明,一阶导数能明显增加3种光谱匹配模型的区分度,尤其是最小距离法增加非常明显.

Tree spectral match models including minimums distance （MD）, spectral angel model （SAM）, spectral correlation fitting （SCF） , were calculated and analyzed based on the spectrums taken from the field. However, the difference of the three models is not significant. The SAM was better but far from the best. Under this circumstance, derivative method was used to enhance the ability of discrimination. The match models were recalculated after the first derivative. Better results are achieved and the discriminability is remarkably enhanced, especially to the minimums distance model.