Kernel eigenmaps based multiscale sparse model for hyperspectral image classification

作者：

Highlights：

• The challenges in high dimensionality of hyperspectral data and its complexity in classification are addressed.

• Spatial–spectral features are extracted using manifold learning.

• Adaptive Sparse classifier with extracted features well exploits the within-class variability.

• Misclassification of similar test pixels is reduced.

摘要

•The challenges in high dimensionality of hyperspectral data and its complexity in classification are addressed.•Spatial–spectral features are extracted using manifold learning.•Adaptive Sparse classifier with extracted features well exploits the within-class variability.•Misclassification of similar test pixels is reduced.

论文关键词：Adaptive sparse representation,Schroedinger eigen maps,Spatial–spectral features,Hyperspectral image classification

论文评审过程：Received 19 January 2021, Revised 28 June 2021, Accepted 2 August 2021, Available online 22 August 2021, Version of Record 3 September 2021.

论文官网地址：https://doi.org/10.1016/j.image.2021.116416