中文摘要：

提出了一种基于主成分分析的高光谱图像波段选择算法。该算法把每个波段被映射到主成分的信息量的大小作为是否被选择的指标，因此，可以保证选择的波段包含原始图像绝大部分信息，而且指标的计算只需要得到原始数据的协方差阵，而不必对原始数据进行真正的主成分变换，极大的降低了计算量。贝叶斯和K-均值分类实验表明．该算法是有效可行的。

英文摘要：

A PCA based band selection algorithm of hyperspectral image is proposed. This algorithm regard the amount of the information that was mapped into the principal components of a given band as the criterion which can be selected, therefore ,it can ensured that the selected bands contain most information of the original data. What＇s more, in order to get the criterion, only the covariance metrics are needed. So, we needn＇t carry principal component transformation on the original data, which save much computation time. The results of the bayesian classification and K-means classification indicate the validity and feasibility of the algorithm.