Improve registration error for advanced super resolution algorithm

The registration error is very important since it affects to final high resolution (HR) image quality. In this paper, we present the improved method to obtain high quality super resolution (SR) reconstruction using auto select input image algorithm. We took the threshold value from maximum motion compensation error (MMCE) to evaluate the suitability of the input image candidates. If the motion compensation error (MCE) between input low resolution (LR) image and reference input LR image is in range of 0 < MCE < MMCE then that input LR image is selected and otherwise that input LR image are neglected. And we selected the reference input LR (SRILR) image by comparing the number of the selected input LR (SILR) images for each reference input LR (RILR) images. Finally, the HR image is generated from restoration and reconstruction block which based on the Hardie algorithm by using SRILR and SILR images. Our proposal algorithm is expected to improve the quality of super resolution (SR) without user participation and it will apply to moving picture.