Asymmetric Dual-Direction Quasi-Recursive Network for Single Hyperspectral Image Super-Resolution

Single hyperspectral image super-resolution aims to reconstruct a high-resolution hyperspectral image from a low-resolution one, which does not use any auxiliary images. For now, existing super-resolution methods often ignore the difference between the features of neighbor and non-neighbor spectral bands, leaving the feature exploration untargeted. As a result, the complementary information of such bands has not been effectively exploited. To do so, we propose an asymmetric dual-direction quasi-recursive network for single hyperspectral image super-resolution, which separately explores the features among neighbor and non-neighbor bands via forward and backward units. By considering the high similarity among neighbor bands, each forward unit thoroughly exploits spatial-spectral features among such bands through two kinds of correspondence aggregation modules. It also preserves a spectral structure by a spectral band grouping strategy and a spatial-spectral consistency module. Owning to the inconsecutive spectra among non-neighbor bands, backward units focus on extracting spatial features in such bands. With the aid of a global feature context fusion module, the information of global non-neighbor context and neighbor bands are adaptively fused, thus improving information completeness and complementarity. Experimental results reported for natural and remote sensing hyperspectral image datasets demonstrate the proposed network not only outperforms the state-of-the-art methods in terms of reconstruction quality and noise suppression, but also requires a smaller memory footprint.

I. Introduction

Hyperspectral imaging has received much attention and made a large number of achievements [1]. Over the past two decades, it is widely used in some practical fields, such as mineral identification [2], plant monitoring [3], and geological exploring [4]. Due to narrow spectral bandwidths in a hyperspectral imaging system, hyperspectral images have a lower spatial resolution than panchromatic, multispectral, and red-green-blue ones. Consequently, they have poor performance in some recognition applications, including change detection [5], anomaly detection [6], and object classification [7], [8], [9]. Therefore, it is essential to develop valid super-resolution methods to enhance their spatial resolution.