A novel spectral-spatial co-training algorithm for the transductive classification of hyperspectral imagery data

作者：

Highlights：

• Applying a transductive learning and co-training to predict the unknown labels of a sparsely labeled hyperspectral image.

• Multi-view learning to manage spectral and spatial views of imagery data, iteratively constructed via collective inference.

• A spatial criterion to select reliably predicted labels and a diversity-class criterion to speed-up the learning process.

• An empirical study including various competitors defined in machine learning and hyperspectral classification.

摘要

Highlights•Applying a transductive learning and co-training to predict the unknown labels of a sparsely labeled hyperspectral image.•Multi-view learning to manage spectral and spatial views of imagery data, iteratively constructed via collective inference.•A spatial criterion to select reliably predicted labels and a diversity-class criterion to speed-up the learning process.•An empirical study including various competitors defined in machine learning and hyperspectral classification.

论文关键词：Hyperspectral imagery classification,Transductive learning,Collective inference,Co-training,Spectral-spatial data

论文评审过程：Received 27 April 2016, Revised 5 September 2016, Accepted 7 October 2016, Available online 8 October 2016, Version of Record 20 October 2016.

论文官网地址：https://doi.org/10.1016/j.patcog.2016.10.010