M3DNet: A manifold-based discriminant feature learning network for hyperspectral imagery

作者：

Highlights：

• A manifold-based discriminant network is proposed for feature learning of HSI.

• An objective function is proposed to maximize the margins among different classes.

• An iterative strategy is designed to optimize the projection matrix for classification.

• The proposed method can improve the performance of many feature extraction methods.

摘要

•A manifold-based discriminant network is proposed for feature learning of HSI.•An objective function is proposed to maximize the margins among different classes.•An iterative strategy is designed to optimize the projection matrix for classification.•The proposed method can improve the performance of many feature extraction methods.

论文关键词：Hyperspectral imagery,Feature extraction,Manifold learning,Graph embedding,Adaptive optimization

论文评审过程：Received 9 August 2019, Revised 13 November 2019, Accepted 14 November 2019, Available online 23 November 2019, Version of Record 29 November 2019.

论文官网地址：https://doi.org/10.1016/j.eswa.2019.113089