Applying sparse based spatial super-resolution for Himawari-8 satellite image
Shiori Ishikuro, Jun Hashimoto, Yuichi Okuyama, Xiang Li · 2019
Super-resolution for meteorological satellite images is expected to improve the accuracy performance in the systems such as elucidations of meteorological, solar irradiance estimation, utilization of photovoltaic power generation, storage batteries and electric vehicles. While various super-resolution studies have been conducted for planimetric features, there seldom exists for meteorological features. In this study, we apply sparse based super-resolution for meteorological satellite images and verify the characteristics of error between original high-resolution image and super-resolution results. We confirmed super-resolution by sparse-based method which keeps the boundary feature of clouds. A sparse constraint gains an advantage for cloud image which contains boundary features from the meteorological satellite. The peak signal-to-noise ratio by the sparse based method was improved 1.43dB at the maximum compared with bicubic interpolation. On the other hand, we show that the sparse-based method still needs further studies to handle the blurry cloud and absent cloud situations.