Estratégias de computação paralela para a restauração de imagens com o funcional de regularização de Tikhonov

• Autor(a) principal: Stutz, Dalmo
• Data de Publicação: 2009
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Biblioteca Digital de Teses e Dissertações da UERJ
• Texto Completo: http://www.bdtd.uerj.br/handle/1/13696
• topic: Image processing - Digital techniques; Parallel processing (Electronic computers); Algorithms; Tikhonov, A. N. (Andrei Nikolaievich), 1906-; Microscope and microscopy; Processamento de imagens - Técnicas digitais; Processamento paralelo (Computadores); Algoritmos; Tikhonov, A. N. (Andrei Nikolaievich), 1906-; Microscópio e microscopia; CNPQ::CIENCIAS EXATAS E DA TERRA::MATEMATICA::MATEMATICA APLICADA

The Atomic Force Microscopy technique allows the nanoscale imaging of almost any type of surface. However, in this scale, the images obtained with this technique can present either poor signal/noise ratios or blurred contents. In order to restore or, at least, decrease the effects of the degradations, image restoration techniques are employed. In the last decades, many techniques have been developed and applied for such purpose, including the restoration technique based on the minimization of Tikhonov's regularization functional, described in this thesis. In recent years, this technique has been used to restore AFM´s image with very satisfactory results. The application of this technique requires, however, a lot of effort and computational power that result on long run-times when processed serially. Moreover, as the electronic equipments have increased their capabilities, the images have increased in resolution as well as the time and the effort for their processing. In the course of time, the processing speedup and the performance improvement have become a critical problem. To solve this, techniques of parallel computation are applied to obtain greater processing speedups. In this thesis, a parallel implementation strategy for the restoration algorithm is described and used to develop a new parallel version of the restoration computational program. The results obtained so far show that the parallel strategy adopted reduced the run times and produced better performances in comparison to the other previous implementations of the same algorithm. Moreover, the new strategy yields higher levels of performance as the restored image resolution increase, making possible the restoration of larger images in time periods relatively shorter.