矿物光谱特征谱段识别方法与应用

Study on hyperspectral mineral identification based on characteristic spectrum peak-valley correlation coefficient method

矿物光谱识别所使用的光谱参量在不同影响因素下的稳定性对矿物识别效果影响很大。基于特征谱段峰谷相关系数法,提出了高光谱矿物识别的新算法,建立了提取稳定光谱参量(峰谷位置)的数学模型和操作流程。该算法通过提取矿物参考光谱峰谷的位置,计算矿物特征谱段的峰和谷与待测矿物光谱相应谱段的相关系数,并以此作为比较矿物光谱相似程度的主要依据。以甘肃北山方山口地区拾金坡金矿床为例,研究并比较新算法与典型算法的蚀变矿物识别。结果表明,本文算法的正确识别率为85%,较典型算法有更好的识别效果。

The stability of the spectrum parameters（width, depth and shape） in a mineral spectrum identification under differentinfluencing factors is greatly influenced by the identification effects. It is shown that the positions of the peak and the valley of differentminerals in the characteristic spectrum are more stable, and they are relatively stable characteristic parameters of the spectrum. This paperproposes a hyperspectral mineral identification algorithm based on the characteristic spectrum peak-valley correlation coefficient method,and the mathematical model and the operation flowchart of extracting the spectral stability parameters（the locations of the peak and thevalley） are established. The algorithm is based on the extraction of the reference spectra peak-valley positions, and the calculations of thepeaks and the valleys of minerals characteristic spectrum and the correlation coefficient of the corresponding measured mineral spectrum, todetermine whether they exceed the thresholds, as the main basis of comparison of the similarity degree of mineral spectra. Gansu BeishanShijinpo gold mining is taken as the study area, using the CAIS/SASI airborne hyperspectral data, and the algorithm is used to identify theregions of alteration minerals, and the results are compared with those obtained with the existing typical algorithms（SFF、SID、SAM）. It isshown that the correct recognition rate of the algorithm is higher, and the accuracy of the algorithm can reach 85%.