Uma metodologia para reconstrução de superfícies a partir de luz estruturada

Abstract

The surfaces reconstruction from two-dimensional images is a problem present in various areas of knowledge, such as reconstruction of organs or parts of the human body from tomography; automatic recognition of characters; computer vision for assembly and inspection, processing of aerial and satellite images for weather forecast, among others. In all these cases it is necessary detailed studies on image processing and the right choice of the technique of reconstruction to be adopted in order to achieve good results. It is presented in this work a model for reconstruction of surfaces based on a technique of three-dimensional reconstruction through structured light. The technique employs a source of straight light focused on the objects of the scene as structured light and applies controlled movement to the objects allowing their surfaces to be swept by that light. Through images captured by a camera containing the projection of the structured light on the objects, the surfaces are reconstructed. The model is validated through a methodology proposed using synthetic images obtained by computer simulation and images obtained by experimental test. Results of the identification of various three-dimensional surfaces are presented as examples of application.