《计算机应用》唯一官方网站 ›› 2022, Vol. 42 ›› Issue (7): 2088-2093.DOI: 10.11772/j.issn.1001-9081.2021050825

• 数据科学与技术 • 上一篇    

基于超像素锚图二重降维的高光谱聚类算法

赖星锦1, 郑致远2, 杜晓颜3, 徐莎1(), 杨晓君1   

  1. 1.广东工业大学 信息工程学院, 广州 510006
    2.火箭军工程大学 第五大队, 西安 710025
    3.96962部队, 北京 102206
  • 收稿日期:2021-05-19 修回日期:2021-09-23 接受日期:2021-09-28 发布日期:2021-09-23 出版日期:2022-07-10
  • 通讯作者: 徐莎
  • 作者简介:赖星锦(1998—),男,广东揭阳人,硕士研究生,主要研究方向:聚类算法
    郑致远(2000—),男,广东广州人,主要研究方向:降维算法
    杜晓颜(1980—),女,河南方城人,工程师,主要研究方向:数据挖掘
    杨晓君(1983—),男,安徽颍上人,副教授,博士,主要研究方向:聚类算法、智能信息处理。
  • 基金资助:
    广东省重点领域研发计划项目(2018B010115001);广东省自然科学基金资助项目(2021A1515011141)

Hyperspectral clustering algorithm by double dimension-reduction based on super-pixel and anchor graph

Xingjin LAI1, Zhiyuan ZHENG2, Xiaoyan DU3, Sha XU1(), Xiaojun YANG1   

  1. 1.School of Information Engineering,Guangdong University of Technology,Guangzhou Guangdong 510006,China
    2.The Fifth Battalion,Rocket Force Engineering University,Xi’an Shaanxi 710025,China
    3.96962 Troops,Beijing 102206,China
  • Received:2021-05-19 Revised:2021-09-23 Accepted:2021-09-28 Online:2021-09-23 Published:2022-07-10
  • Contact: Sha XU
  • About author:LAI Xingjin, born in 1998, M. S. candidate. His research interests include clustering algorithm.
    ZHENG Zhiyuan, born in 2000. His research interests include dimension reduction algorithm.
    DU Xiaoyan, born in 1980, engineer. Her research interests include data mining.
    YANG Xiaojun, born in 1983, Ph. D., associate professor. His research interests include clustering algorithm, intelligent information processing.
  • Supported by:
    Research and Development Program in Key Areas of Guangdong Province(2018B010115001);Natural Science Foundation of Guangdong Province(2021A1515011141)

摘要:

针对传统谱聚类算法难以应用于大规模高光谱图像,以及现有的改进谱聚类算法对大规模高光谱图像的处理效果不佳的问题,为降低聚类数据的复杂度,以降低聚类过程的计算成本从而多方面提升聚类性能,提出一种基于超像素锚图二重降维的高光谱聚类算法。首先,对高光谱数据进行主成分分析(PCA)处理,并针对高光谱图像的区域特性对其进行基于超像素切割的降维;其次,通过构造锚图的思想对上一步所得数据进行锚点的选取,并构建邻接锚图来实现二重降维,从而进行谱聚类;同时,为去除算法运行中人为调节参数的环节,在构建锚图时采用一种去除高斯核的无核锚图构造方式以实现自动构图。在Indian Pines数据集和Salinas数据集上的实验结果表明所提算法在保证可用性与低耗时的前提下可提高聚类的整体效果,从而验证了所提算法能提高聚类的质量与性能。

关键词: 高光谱图像, 超像素切割, 锚图, 谱聚类, 降维

Abstract:

Traditional spectral clustering algorithms are difficult to be applied to large-scale hyperspectral images, and the existing improved spectral clustering algorithms are not effective in processing large-scale hyperspectral images. To address these problems, a hyperspectral clustering algorithm based on double dimension-reduction of super-pixel and anchor graph was proposed to reduce the complexity of clustering data that is to reduce the computational cost of clustering process, thereby improving the clustering performance in many aspects. Firstly, Principal Component Analysis (PCA) was performed to the hyperspectral image data, and dimension-reduction was carried out to the data based on super-pixel segmentation according to the regional characteristics of hyperspectral image. Then, the anchor points of the data obtained in previous step were selected with the idea of constructing anchor graph. And the adjacent anchor graph was constructed to achieve double dimension-reduction for spectral clustering. At the same time, in order to remove the artificial adjustment of parameters in the operation of the algorithm, a kernel-free anchor graph construction method with the Gaussian kernel removed was used in the construction of anchor graph to achieve automatic graph construction. Experimental results on Indian Pines dataset and Salinas dataset show that the proposed algorithm can improve the overall effects of clustering with guaranteeing availability and low time consumption, thus verifying that the proposed algorithm can improve the quality and performance of clustering.

Key words: hyperspectral image, super-pixel segmentation, anchor graph, spectral clustering, dimension-reduction

中图分类号: