| 29-10-2021 |
Multiresolution analysis and superresolution
Multiresolution analysis consists in constructing a filtration of L2 of closed subspaces Vj such that each one represents functions at a scale 2j. The ortogonal projection onto Vj represents the approximation at scale 2j whereas the difference between the projections onto Vj and Vj+1 represents the details at scale 2j. A typical signal distortion process consists in removing structure at small scales. This is modeled through convolutions and resampling. Is it possible to leverage multiresolution analysis to recover the missing details? In this case we do not want to solve the problem for any function, thus constraining the worst-case, but only for those that are of interest and have been distorted by the process under study. The goal of this iPre is to review the existing literature connecting multiresolution analysis with this problem, and to propose a mathematical model that would allow us to answer this question.
Keywords:
análisis de fourier
superresolución
Prerequisites:
IMT2113
Evaluation method: Nota 1-7, with 0/1 available vacants |
Mentor(s): Open in the plataform |