Hyperspectral Image Classification Based on Multi-Scale Residual Network with Attention Mechanism

In recent years, image classification on hyperspectral imagery utilizing deep learning algorithms has attained good results. Thus, spurred by that finding and to further improve the accuracy, we propose a multi-scale residual convolutional neural network model fused with an efficient channel attention (MRA-NET) is appropriate for classification. The suggested technique comprises multi-staged architecture, where initially spectral information of reduced into two-dimensional tensor, principal component analysis (PCA) scheme. Then, constructed low-dimensional input our proposed ECA-NET network, which exploits advantages its core components, i.e., structure mechanisms. We evaluate performance MRA-NET three public available datasets demonstrate that, overall, accuracy method 99.82 %, 99.81%, 99.37, respectively, higher compared corresponding current networks such as 3D (CNN), three-dimensional convolution (RES-3D-CNN), space–spectrum joint (SSRN).