针对红外高光谱数据目标鉴别问题, 根据目标光谱特点, 采用均匀区域法进行噪声评估、PCA和LDA算法进行数据降维与特征提取、光谱间最小距离匹配算法进行数据分类这三个步骤, 对高光谱数据进行分析处理。重点对PCA/ICA/LDA算法及LDA算法的性能进行了分析对比, 实现了不同目标的红外光谱鉴别。根据对比结果可以看出, LDA算法在光谱数据特征分离方面, 与PCA和ICA两个算法对比具有较好的效果。

According to the problem of infrared hyperspectral data identification and the characteristics of target spectrum, hyperspectral data are analyzed and processed through three steps, first to noise estimation by homogeneous area method, second to data dimension reduction and feature extraction by principal component analysis (PCA) and linear discriminant analysis (LDA) algorithms and third to data classification by spectral minimum distance matching algorithm. The characteristics of PCA, independent component analysis (ICA) and LDA algorithms are compared and analyzed to realize infrared spectrum identification of different targets. Compared results show that LDA algorithm has better effect on spectral data characteristic separation comparing with that of PCA and ICA algorithms.