基于双边滤波和空间邻域信息的高光谱图像分类方法

发布时间：2018-02-28 19:19

本文关键词： 高光谱图像分类双边滤波像元邻域信息空间相关信息　出处：《农业机械学报》2017年08期 　论文类型：期刊论文

【摘要】：提出了一种基于双边滤波和像元邻域信息的高光谱图像分类(BS-SVM)算法。该方法首先利用双边滤波器提取经主成分分析降维后的高光谱图像空间纹理信息,然后通过设计一种高光谱像元邻域信息来构建高光谱的空间相关信息,最后将2种空间信息融合后与光谱信息结合,形成空谱信息(空间信息和光谱信息)后交由支持向量机完成分类。实验结果表明,相比单纯使用光谱信息的支持向量机的分类方法以及基于Gabor滤波的空谱信息结合分类方法,所提出的BS-SVM方法分类精度有较大幅度提高,充分证明了该方法的有效性。
[Abstract]:A BS-SVM algorithm for hyperspectral image classification based on bilateral filtering and pixel neighborhood information is proposed. Firstly, the spatial texture information of hyperspectral image reduced by principal component analysis (PCA) is extracted by two-sided filter. Then the hyperspectral spatial correlation information is constructed by designing a hyperspectral pixel neighborhood information. Finally, the two spatial information are fused and combined with the spectral information. Spatial spectrum information (spatial information and spectral information) is formed and classified by support vector machine. The experimental results show that compared with the classification method of support vector machine using spectral information only and the method of combining space spectrum information based on Gabor filter, the experimental results show that, The classification accuracy of the proposed BS-SVM method is greatly improved, which fully proves the effectiveness of the proposed method.
【作者单位】：广东交通职业技术学院计算机工程学院;哈尔滨工程大学信息与通信工程学院;青岛理工大学通信与电子工程学院;
【基金】：国家自然科学基金项目(61275010、61675051) 国家星火计划项目(2014GA780056) 广东交通职业技术学院校改重点科研课题(2017-1-001)
【分类号】：TN713;TP751

【相似文献】