Abstract: The deep network can effectively improve the accuracy of the reconstructed image, but it has a large number of parameters, which makes the training time too long. Therefore, this paper improves the super-resolution reconstruction algorithm of remote sensing image based on recursive residual network. The global residual learning and local residual learning are combined to effectively reduce the difficulty of training deep network and control the network parameters through recursive learning. The experimental results show that the recursive residual network is effective in the super-resolution reconstruction of the remote sensing image, and the improved network can obtain better subjective visual effect and objective evaluation index.

Key words: recursive residual network, remote sensing image super-resolution reconstruction, residual learning, recursive learning

摘要： 深层网络有效地提高了重建图像的精度，但是拥有大量参数，使训练时间过长。因此，改进了一种基于递归残差网络的遥感图像超分辨率重建算法，将全局残差学习和局部残差学习相结合，有效地降低训练深层网络的难度，并且通过递归学习控制网络参数。实验结果证明了递归残差网络在遥感图像超分辨率重建中的有效性，改进的网络可以获得更好的主观视觉效果以及客观评价指标。

关键词: 递归残差网络, 遥感图像超分辨率重建, 残差学习, 递归学习