Detalhes bibliográficos
| Ano de defesa: |
2007 |
| Autor(a) principal: |
Martins, Ana Luísa Dine |
| Orientador(a): |
Mascarenhas, Nelson Delfino d'Ávila
 |
| Banca de defesa: |
Não Informado pela instituição |
| Tipo de documento: |
Dissertação
|
| Tipo de acesso: |
Acesso aberto |
| Idioma: |
por |
| Instituição de defesa: |
Universidade Federal de São Carlos
|
| Programa de Pós-Graduação: |
Programa de Pós-Graduação em Ciência da Computação - PPGCC
|
| Departamento: |
Não Informado pela instituição
|
| País: |
BR
|
| Palavras-chave em Português: |
|
| Área do conhecimento CNPq: |
|
| Link de acesso: |
https://repositorio.ufscar.br/handle/20.500.14289/349
|
Resumo: |
Super resolution image reconstruction consists in using a set of low resolution images from the same scene to generate a high resolution estimate of the original scene. For that purpose, all the observed low resolution images need to have sub-pixel displacements among each other. In this way, there is more than just the same information replicated in each image and then the uncertainty inherent to the displacements can be used as additional information to increase the spatial resolution. This master s thesis proposes a Bayesian approach for the super resolution reconstruction problem using Markov Random Fields and the Potts-Straus model for the image characterization. Therefore, it is possible to incorporate previously known context spatial information about the high resolution image to be estimated. Moreover, a discontinuity adaptive ICM algorithm was used to estimate the maximum a posteriori solution. Using an initial high resolution estimate constructed from the registration and interpolation of all the observations made it possible to reconstruct an image that respected the initially presented discontinuities. We also observed that the resulted high resolution image hold finner details when compared to the initial estimation. |
