基于波段子集的独立分量分析的特征提取的高光谱遥感影像分类

针对高光谱影像数据具有波段众多、数据量较大的特点,本文提出了一种基于波段子集的独立分量分析(ICA)特征提取的高光谱遥感影像分类的新方法。以北京昌平小汤山地区的高光谱影像为例,根据高光谱遥感影像的相邻波段的相关性进行子空间划分,在各个波段子集上采用ICA算法进行特征提取,将各个子空间提取的特征合并组成特征向量,采用支持向量机(SVM)分类器进行分类。结果表明:该方法分类精度最佳(分类精度89.04%,Kappa系数0.8605,明显优于其它特征提取方法的SVM分类,有效地提高了高光谱数据的分类精度。

Classification Technique for Hyperspectral Image Based on Bands Subspace of ICA Feature Extraction and SVM

The paper proposes a hyperspectral remote sensing image classification algorithm based on SVM(Support Vector Machine).The SVM uses the features extracted from subspace of bands(SOB) of ICA(independent component analysis).Taking hyperspectral data(PHI sensor getting 80 bands) data in Beijing Changping Xiaotangshan area for example,the adjacent band correlation in each band of the hyper spectral image is used in order to divide the feature space into several SOBs.In the SOBs we use ICA algorithms?for feature extraction.Then we combine the extracted features into the feature vector for classification.it is found that the feature extraction algorithm proposed by the paper with accuracy 89.04% and kappa coefficient 0.8605 has the best classification result,better than the results of four kinds of conventional feature extraction algorithms.The study also indicated that the method proposed by this paper optimizes spectral information can effectively improve the classification precision of hyperspectral data.