【摘要】 本文给出2021年“高教社杯”全国大学生数学建模竞赛E题“中药材鉴别”可行的解法.本题基于光谱特征对中药材种类和产地进行鉴别，提供了非监督识别、监督识别和半监督识别3种不同场景.针对光谱特征之间强关联的特点，提出了分段进行特征提取的思路，通过系统比较发现，分段主成分分析或者分段多项式拟合进行特征提取再结合线性判别分析可以得到比较好的预测效果.更多还原

【Abstract】 This paper presents a feasible solution to problem E "Herb Discrimination" in 2021 Higher Education Press Cup China Undergraduate Mathematical Contest(CUMCM 2021). The aim of this problem is to determine the class and original place for herbs based on their infrared spectrums, which provide scenarios for unsupervised learning, supervised learning and semi-supervised learning. Considering the continuous characteristic of spectrum features, i.e, the strong correlation among the adjacent spectrum features, we propose the idea of block-wise feature extraction. With systematic comparisons, we find block-wise PCA or block-wise polynomial fitting, combined with linear discrimination analysis(LDA) can achieve better performance.更多还原