JOINTIMAGE REGISTRATIONAND SUPER-RESOLUTIONUSINGNONLINEAR LEAST SQUARESMETHOD
Lists
Only you see this
Abstract
Thispaperproposes anewalgorithm tointegrate image registration intoimagesuper-resolution (SR)byfusing multiple blurred low-resolution (LR)images torender a high-resolution (HR)image. Conventional super-resolution (SR)imagereconstruction algorithms assumeeither the estimated motion(displacement) errorsby existing registration methods arenegligible orthedisplacement is knownapriori. Thisassumption, however, isimpractical as theperformance ofexisting registration algorithms isstill lessthanperfect. Inviewofthis, we present a new estimation framework that performs joint image registration andHR reconstruction. An iterative schemebasedon nonlinear least squares method isdeveloped toestimate the motion shift (displacement) andHR imageprogressively. Themotion modelthat isconsidered inthis workincludes bothtranslation aswellasrotation. Experimental results showthat theproposed methodiseffective inperforming image super-resolution. Index Terms- Imagesuper-resolution, image
BibTeX
@inproceedings{yap2007jointimage,
author = {Kim–Hui Yap},
title = {{JOINTIMAGE REGISTRATIONAND SUPER-RESOLUTIONUSINGNONLINEAR LEAST SQUARESMETHOD}},
booktitle = {IEEE International Conference on Acoustics, Speech and Signal Processing},
year = {2007},
doi = {10.1109/icassp.2007.365969},
}Loading related content...