Refinamento de Reconstrução 3D através de Seleção Apropriada de Keyframes

Andréa SantosMárcio BuenoAndré FerreiraJudith KelnerIony Siqueira

This paper introduces a measure of preprocessingthat aims to achieve a set of images (keyframes),carefully selected, initially captured from a real-worldcamera, to be later used in the reconstruction of 3D models.More precisely, it is a technique that evaluates the quality ofthe captured frames, allowing that blurring as Motion Blurcould be identified before the 3D modeling processes starts,and those which have the best sharpness levels will be usedas representatives of the sequence images. So, using thissimple and effective metric the computational cost in the 3Dreconstruction can be reduced. It is important to considerthat this metric is robust enough to identify in a particularsequence of images, different scenarios, i.e., apply newstandards to calculate the presence or absence of MotionBlur according to the level of contrast of each scenario. Theresults will be compared to other metrics and will be appliedin the reconstruction of 3D models for better visual analysisof the final result.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: