Parallel Repairing 3D Fuzzy Images into Well-Composed Images

• Autor(a) principal: Germano, Rafael Lucena
• Data de Publicação: 2018
• Tipo de documento: Trabalho de conclusão de curso
• Idioma: eng
• Título da fonte: Repositório Institucional da UFRN
• Texto Completo: https://repositorio.ufrn.br/handle/123456789/34194
• topic: well-composedness

This work presents a parallelized version in Compute Unified Device Architecture of the algorithm presented in (SIQUEIRA et al., 2008) for repairing images into well-composed ones, as well as a comparison between heuristics to obtain the well-composed image which minimizes the difference between the generated well-composed image and the original image. The algorithm is based on successively changing the points from one object to another until the image becomes well-composed. Well-composed images are images on which the intersection of the voxels of an object with its complement forms a topological surface. Such images enjoy very useful properties which reduce the processing time of algorithms, such as thinning and surface extraction algorithms. Lastly, the performance of the parallel and sequential versions are compared and an analysis of the produced well-composed images is done.