Cloud Image Super-Resolution Based on Residual Network
Guodong Jing, Yun Ge · 2021 7th International Conference on Computer and Communications (ICCC) · 2021
Using cloud image to make weather forecast is the main method for meteorological research in recent years. However the cloud images obtained from current equipment have the problem of low resolution and high noise. To overcome these problems and get a high resolution result, we proposed a cloud image super-resolution method based on residual convolution network. The method can automatically learning the appropriate upscale parameters and merging parameters, which can avoid artificial impacting. In addition to that, the residual structure can help us to avoid gradient disappeared problem. In this method, the feature learning procedure and the reconstruction method was proposed. Experiments show that this method can effectively improve the accuracy of the reconstructed image in detail.