基于谱库的匹配识别是应用拉曼光谱进行物质成分鉴别的关键, 会直接影响匹配结果的准确性。在谱库匹配中, 尤其是针对混合物的光谱, 利用单一的匹配特征无法全面反映被测样本光谱与谱库光谱的相似性, 光谱匹配识别时需要综合考虑多种匹配特征。采用逻辑回归数学模型融合谱峰匹配系数、非负最小二乘匹配系数以及夹角余弦匹配系数, 提出了一种新的光谱集成匹配方法。该方法既考虑了被测样品的谱峰信息, 又考虑了其全谱信息。基于20种氨基酸混合物拉曼光谱谱库匹配的实验结果表明：所提光谱集成匹配方法具有更低的误判率。

Library-based matching recognition is the key to the application of Raman spectroscopy for material composition identification, which directly affects the accuracy of matching results. For the library matching, especially for the mixture spectrum, the single matching feature can not fully reflect the similarity between the measured sample spectrum and the spectral spectrum in the library. The spectral matching needs to comprehensively consider multiple matching features. In this paper, a new spectral integration matching method is proposed by using the logistic regression mathematical model to fuse the peak matching coefficient, the non-negative least squares matching coefficient and the cosine matching coefficient. The new method takes into account both the spectral peak information and full spectrum information of the sample. The experiment results based on the Raman spectroscopy library matching of 20 kinds of amino acid mixtures show that the spectral integration matching method has lower false positive rate.