In the present paper, the similar spectra of 18 samples, which include Astragalus, red-blue Astragalus and Codonop-sis, were obtained in the range of 1 600~700 cm‐1. The result showed that all kinds of herbs have their own characteristic similar spectra, and 18 samples can be identified according to the characteristic similar spectra. Furthermore, three correlation coefficients of 93 ganoderma samples were calculated which is in the range of 1 560~l 502, 1 460~l 421 and 1 319~1 260 cm‐1 according to the information of similar spectrum of infrared spectrum of ganoderma. Without priori knowledge of the classification of these samples,the K-means cluster analysis can successfully divide them into four classes, i. e. Ganoderma lucidum and Ganoderma sinensis, Ganoderma atrum , Ganoderma aoshiba , Ganoderma multiplicatum. This result is consistent with the result of morphological classification.%红外光谱相似谱能描述两红外光谱相似的程度,在波数-νi处相关谱的值ri能反映两红外光谱在[-νi-n,-νi+n]区间内线性相关程度的大小,值在-1～1之间,ri＞0意味着两光谱正相关,ri＜0意味着负相关,ri=0则没有线性关系.在偏移量n一定的条件下,依相似谱形状的差异可辅助红外光谱的解析,用相似谱或相似谱提供的信息提取的光谱特征值可用于模式识别.计算了18个黄芪、红兰芪和党参样本在1 600～700 cm-1范围内的红外光谱相似谱,结果表明各类药材均有各自的特征相似谱,凭借这些可实现对它们的鉴别；依灵芝红外光谱相似谱提供的信息,计算了1 560～1 502,1 460～1 421和1 319～1 260 cm-1三个波数区间内93个灵芝样本红外光谱的相关系数,在无先验知识的情况下,K-means聚类分析法成功地将其分为了赤芝和紫芝类、黑芝类、青芝类和黄芝类,这一分类结果与形态分类法得到的结果基本一致.
展开▼
