Hyperspectral image classification using an extended Auto-Encoder method
作者：
Highlights：
• A modified Auto-Encoder based on MM technique is proposed for Hyper-Spectral Image classification.
• Three main modifications are made. First, SAM is used as a regularization term to construct the weights of the Auto-Encoder.
• Second, Fuzzy weighting is used to fine-tune the parameters.
• Third, multi-scale features (MSF) are used to improve the performance of the Auto-encoder resulting in the proposed method - MSF-EAEMM.
• MSF-EAEMM achieves high accuracy and reduces the time complexity using lower orientation and scales of Gabor filter.
摘要
•A modified Auto-Encoder based on MM technique is proposed for Hyper-Spectral Image classification.•Three main modifications are made. First, SAM is used as a regularization term to construct the weights of the Auto-Encoder.•Second, Fuzzy weighting is used to fine-tune the parameters.•Third, multi-scale features (MSF) are used to improve the performance of the Auto-encoder resulting in the proposed method - MSF-EAEMM.•MSF-EAEMM achieves high accuracy and reduces the time complexity using lower orientation and scales of Gabor filter.
论文关键词：Hyperspectral image,Image classification,Auto-Encoder
论文评审过程：Received 30 April 2020, Revised 13 October 2020, Accepted 17 December 2020, Available online 26 December 2020, Version of Record 4 January 2021.
论文官网地址：https://doi.org/10.1016/j.image.2020.116111
