Analysis of shift estimation techniques of super resolution applied to satellite images

Abstract

There is a high demand for high-resolution satellite sensing in modern application. Super Resolution (SR) offers an affordable solution for this high demand. The accuracy of super resolution depends on the accuracy of determining the difference between the low-resolution images. Shift estimation is the first and the most critical step in super resolution. This paper discusses shift estimation techniques in both spatial and frequency domains. It compares Vandewalle algorithm, Lucchese algorithm and Keran algorithm. Two real satellite images (SPOT-5) are used in the experiment. The images have -0.5 and 0.5 sub pixel shift in the horizontal and vertical directions respectively. The experimental results show that the Estimation shift parameters in spatial domain methods outperform the frequency domain methods.