论文部分内容阅读
传统高光谱遥感影像逐像素分类方法未考虑像元之间的空间关联性且泛化性能较低。形态学属性剖面是表征影像空间结构的有效方法,同时集成学习可显著提升分类算法的泛化能力。为了在高光谱影像分类中充分利用影像的空间信息并提高分类的稳定性,提出一种基于形态学属性剖面高光谱遥感影像集成学习分类方法。首先,用主成分分析和最小噪声变换进行特征提取,并借助形态学属性剖面获取影像的多重空间特征;然后用极限学习和支持向量机的方法进行分类;最后将多个分类结果以多数投票的方式集成。区别于已有集成学习方法,综合考虑了不同特征提取和不同分类方法的联合集成,并将形态学属性剖面引入其中以充分利用影像的空间信息。采用AVIRIS和ROSIS两组高光谱数据检验该方法的分类性能,实验结果表明该方法可获得高精度和高稳定性的分类结果,总体精度分别达到83.41%和95.14%。
The traditional pixel-by-pixel hyperspectral classification method does not consider the spatial correlation between pixels and the generalization performance is low. Morphological attribute profile is an effective method to characterize the spatial structure of image, while integrated learning can significantly improve the generalization ability of classification algorithm. In order to make full use of the spatial information of images and improve the stability of classification in hyperspectral image classification, an integrated learning classification method based on morphological attributes of hyperspectral remote sensing images is proposed. Firstly, the feature extraction is carried out by principal component analysis and minimum noise transform, and the multi-spatial features of the image are obtained by morphological attribute profiles. Then the classification is carried out by the methods of limit learning and support vector machines. Finally, Way integration. Different from existing integrated learning methods, this paper considers the combination of different feature extraction and different classification methods, and introduces morphological attribute profiles into it to take full advantage of the spatial information of images. The AVIRIS and ROSIS hyperspectral data were used to test the classification performance of the proposed method. The experimental results show that the proposed method can obtain the classification results with high precision and high stability, with the overall accuracy reaching 83.41% and 95.14% respectively.