Une approche multi-échelle de super-résolution à partir de séquences d'images

The reconstruction of a high resolution image from a low resolution image sequence using motion information is considered. This issue is formulated as an inverse problem and is solved using an iterative method called generalized Kaczmarcz algorithm. We describe a low computational cost implementation of the method based on the estimation of a 2D parametric motion model. We also propose two improvements to this algorithm. The rst one is the introduction of a motion detection map to deal with areas where the global motion model is not valid. The second one is a temporal multiscale approach allowing us to improve motion estimation within the super-resolution process.