Rank-Aware Generative Adversarial Network for Hyperspectral Band Selection

Traditional clustering-based band selection (BS) methods treat each band as individuals, and selection is conducted by enlarging the difference between clusters, which leads to the loss of band interaction and information saliency evaluation. In this article, we propose a BS method named rank-aware generative adversarial network (R-GAN) to address these problems. First, centralized reference feature extraction (FE) with GAN aids R-GAN to combine interpretability and interband relevance. Then, the reference feature is refined with the saliency estimation provided by the rank-aware strategy. According to data characteristics, there are two versions of rank computation including tensor and matrix. Finally, the structural similarity index measurement (SSIM) maps the saliency to the original data space to obtain the final BS result. Extensive comparison experiments with popular existing BS approaches on five hyperspectral images (HSIs) datasets show that the proposed R-GAN can address spectral saliency effectively and select more informative band subsets, which outperforms other competitors for both detection and classification tasks. For example, on the SD-1 dataset, the ten bands selected by R-GAN achieve 0.982 +/- 0.003 with an improvement of 13.7% in the area under the curve (AUC) value of anomaly detection performance. The peaked accuracy surpasses the baseline by 0.46% for the classification on the PaviaU dataset.

Automated summary

This article proposes a band selection method called rank-aware generative adversarial network (R-GAN) to address the issues of band interaction and information saliency evaluation in traditional clustering-based methods. The proposed R-GAN combines interpretability and interband relevance through centralized reference feature extraction with GAN, and refines the reference feature with a saliency estimation strategy. Experimental results demonstrate that R-GAN can effectively address spectral saliency and select more informative band subsets, outperforming other competitors in detection and classification tasks.