中文摘要：

采用ASD公司生产的FieldSpec HandHeldTM地物光谱仪,分别于2005、2006、2008年冬季跟踪观测杉木、马尾松、黑松、雪松等针叶树种的高光谱数据,经筛选后获取有效观测数据160条,其中120条作为训练集,40条作为测试集。将平滑去噪的一阶微分高光谱数据进行PCA方法和GA方法降维,然后利用BP神经网络和支持向量机（SVM）对降维后的测试集数据进行分类。结果表明：PCA—BP神经网络模型分类准确率95%,PCA—SVM分类准确率97.5%,GA和BP分类准确率92.5%,GA-SVM分类准确率100%。这说明两种降维方式结合支持向量机的分类均优于其与BP神经网络结合的分类,基于GA的降维方法对高光谱波段的选择更有效率,具有较好的应用前景。

英文摘要：

The hyper-spectra data of coniferous species,including Cunninghamia lanceolata,Pinus massoniana,Pinus thunbergii and Cedrus deodara,were collected using the FieldSpec HandheldTM of ASD Inc.in Winter 2005,2006 and 2008 respectively,and the efficient data of 160 samples were obtained through screening.All samples were divided randomly into two groups,one group with the 120 samples used as the calibrated set,and the other with the 40 samples as the calibrated set.The samples data were pretreated by the methods of smoothing and first derivative.The pretreated data were analyzed by Principal Component Analysis（PCA） and Genetic Algorithms（GA） methods,and then which were classified by Back Propagation（BP） and Support Vector Machine（SVM）.The results show： the accuracy was 95% by PCA-BP neural network model,97.5% by PCA-SVM model,92.5% by GA-BP model,and 100% by GA-SVM model.The results show that the classification using the two methods combined with SVM does much better than that combined with BP,the selection of hyper-spectral band base on GA method is more efficient way,and which has a good foreground in the application.