Remote Sensing Image Super-Resolution using Multi-Scale Convolutional Neural Network

Remote sensing images have advantages in large-area imaging and macroscopic integrity. However, in most commercial applications, further recognition and processing becomes difficult due to the low spatial resolution of the acquired images. Therefore, improving the resolution of remote sensing images has important practical significance. To solve this problem, we propose a remote sensing image super-resolution method based on deep learning technology. In order to obtain more detailed image information, we introduce multi-scale convolution to implement feature extraction and deconvolution be used to achieve the final 3× image reconstruction without bicubic interpolation. Experimental results show that our network achieves better performance than prior art methods and visual improvement of our results is easily noticeable.