Preprint serie B95/2017
RECONSTRUCAO ADAPTATIVA DE SUPERFICIES IMPLICITAS A PARTIR DE IMAGENS DE PROFUNDIDADE
Hallison Oliveira da Paz
Keywords: Depth-Image, implicit surfaces, Geometric Reconstruction

This work constitutes a study about the problem of implicit surfaces reconstruction from depth images data, taking into account the current scenario of mobile devices. We propose a method for obtaining 3D models with a geometry adapted to the resolution of the represented surface as opposed to the resolution of the captured image. We address the acquisition of depth images in a mobile device from different points of view, the data registration in relation to a frame of reference in the Euclidean space, as well as the fusion of these data through the construction of an extrinsic representation of the surface which allows the reconstruction of the captured object or scene incrementally. In this work, the experiments and computation were performed on a regular computer, but the choices of solutions and algorithms are designed to test the viability of a method whose complete pipeline can run on a mobile device. We show the advantages from using the features available on a current mobile device in designing solutions to classic problems of the area. On the other hand, we also explore computationally efficient methods to operate according to the hardware limitation these platforms have over an ordinary computer. 

MSC 2000: 68U05