Scale conversion of multi sensor remote sensing image using single frame super resolution technology

Despite its high spatial resolution (15m), the 60km swath width of Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) can limit the effectiveness of accurately large-scale study. One possible solution is to down scale low spatial resolution but width swath satellite image data (e.g. Landsat TM/ETM+ with spatial resolution 30m and swath width 185km). This paper proposed a method related with super resolution named support vector regression (SVR) to convert the low resolution ETM+ image to a high resolution ASTER image. The experiments are conducted on the subset of the ETM+ scene and ASTER UV/NIR scene. The predicted results show that the proposed method is better than the interpolation method common adopted in down scale, both visually and objectively. Thus it can be used to make multi sensor and multi resolution analysis, even has the potential to extend the ASTER scene´s swath width to the same with the ETM+´s.