Revisiting Deep Hyperspectral Feature Extraction Networks via Gradient Centralized Convolution

The hyperspectral images are composed of a variety of textures across the different bands which increase the spectral similarity and make it difficult to predict the pixel-wise labels without inducing additional complexity at the feature level. To extract robust and discriminative features from the different regions of land cover, the hyperspectral research community is still seeking such type of convolutions which can efficiently deal with fine-grained texture information during the feature extraction phase, which often overlook this aspect by vanilla convolution. To overcome the above shortcoming, this article proposes a generalized gradient centralized 3D convolution (G2C-Conv3D) operation, which is a weighted combination between the vanilla and gradient centralized 3D convolutions (GC-Conv3D) to extract both the intensity-level semantic information and gradient-level information. This can be easily plugged into the existing HSI feature extraction networks to boost the performance of accurate prediction for land-cover types. To validate the feasibility of the proposed G2C-Conv3D, we have considered the existing CNN3D, MS3DNet, ContextNet, and SSRN feature extraction models and as well as CAE3D, VAE3D, and SAE3D autoencoder (AE) networks, respectively. All these networks are embedded with G2C-Conv3D convolution to implement both generalized gradient centralized feature extraction networks (G2C-FE) and generalized gradient centralized AE networks (G2C-AE) for fine-grained spectral–spatial feature learning. In addition, G2C-Conv2D is also considered with few networks. The extensive experiments are conducted on four most widely used hyperspectral datasets i.e., IP, KSC, UH, and UP, respectively, and compared with the nine methods. The results demonstrate that the proposed G2C-Conv3D can effectively enhance the feature learning ability of the existing networks and both the qualitative and quantitative results show the superiority and effectiveness of the proposed G2C-Conv3D. The source codes will be publicly available at https://github.com/danfenghong/G2C-Conv3D-HSI.

I. Introduction

With the rapid development of remote-sensing satellite sensors, the land-cover classification with various earth observation (EO) datasets, e.g. hyperspectral images (HSI) [1], synthetic aperture radar (SAR) [2], and light detection and ranging (LiDAR) [3] are becoming the hot research topic among the remote-sensing communities [4]. Hyperspectral images contain rich spectral information which is encoded in the form of spectral bands and can be used to distinguish between the different land-cover classes of interest but especially having slightly more complex spectral behaviors [5]. On the other hand, the distinction between the spectrally similar texture classes can be made easily by capturing more subtle discrepancies from the contiguous representation of spectral bands associated with each pixel. However, due to the high spectral variability causes by the noise artifacts [6], [7], the discriminative feature extraction becomes difficult from such data.