Integrating Convolutional Neural Network and Gated Recurrent Unit for Hyperspectral Image Spectral-Spatial Classification.

In this paper, we propose a novel deep learning framework for hyperspectral image (HSI) spectral-spatial classification. This framework mainly consists of two components: convolutional neural network (CNN) and gated recurrent unit (GRU). CNN is used to automatically extract the high-level spatial features of each band, which are then fed into a fusion network based on GRUs. This fusion network combines feature-level fusion and decision-level fusion together in an end-to-end manner, thus sufficiently fusing the complementary information from different spectral bands. To demonstrate the effectiveness of the proposed method, we compare it with several state-of-the-art deep learning methods on two real HSIs. Experimental results show that the proposed method can achieve better performance than comparison methods.