Affinity Propagation for Hyperspectral Band Selection

2nd International Symposium on Resource Exploration and Environmental Science

生态环境科学

Ren, Zhiwei^1 ; Wu, Lingda^2

Department of Graduate Management, Space Engineering University, Beijing

101416, China^1

Science and Technology on Complex Electronic System Simulation Laboratory, Space Engineering University, Beijing

101416, China^2

关键词: Affinity propagation;    Affinity propagation clustering;    Calculation of similarities;    Classification accuracy;    Clustering methods;    Dimensionality reduction;    Information redundancies;    Minimum distance classifiers;   

Others  :  https://iopscience.iop.org/article/10.1088/1755-1315/170/2/022061/pdf
DOI  :  10.1088/1755-1315/170/2/022061

学科分类：环境科学（综合）

来源: IOP

【 摘 要 】

Because hyperspectral images have the characteristics of high correlation between bands and strong information redundancy, the reduction in dimension of hyperspectral images is an important step in the pre-processing of hyperspectral images. Band selection can preserve the physical meaning of the original data while reducing dimension and has application in many aspects. Affinity Propagation Clustering (AP) is a clustering method proposed by Fray et al. in 2007. AP clusters based on the correlation between data points and treats all data points as potential cluster centers. This paper proposes a band selection method based on AP clustering, which introduces wavelet transform into the calculation of similarity and preference value in clustering algorithm. The dimensionality reduction results are input into the minimum distance classifier for classification, and the classification accuracy was calculated. The dataset is validated by the Indiana Pines dataset. The experimental results verify the effectiveness of the proposed method.