A NURBS based 3D object reconstruction using colour-coded structured light

This thesis introduces a new robust and high precision methodology and corresponding techniques for reconstructing three-dimensional (3-D) views of real objects from one single two-dimensional (2-D) image. The reconstruction process relies on obtaining the two dimensional image of a real object from a camera, on which a colour coded structured light is projected. The method does not require any a-priori knowledge of the absolute positioning or orientation of the camera and the projector which illuminates the scene with the colour coded structured light. Prior to the 3-D reconstruction steps, a calibration process is used to provide a high precision calculation of both the camera's and the projector's intrinsic and extrinsic parameters. These parameters are essential to the 3-D object reconstruction technique as introduced in this thesis. The structured light is used to determine unique patterns on the object's surface. The lines provide a series of control points which once extracted from the object's 2-D view, are used in the creation of Non-Uniform Rational B-Spline (NURBS) curves in 2-D. Those NURBS curves are then projected into space to eventually re-create a 3-D surface. The technique's precision depends on the structured light sampling rate adaptation on the object surface. In a recent test, a 7 tau (thousands of an inch) precision was achieved with a relatively smooth object.; The thesis will focus on the 3-D reconstruction of an object based on its single 2-D view. The approach proposed is end-to-end since it handles all the steps necessary to go from a 2-D view of an object to its modelization using NURBS curves and eventually to a surface in 3-D which can be manipulated with 3-D editing software on a computer monitor. The thesis's results are applicable to many domains. One of those domains is the CAD/SLA systems for manufacturing applications. Furthermore, NURBS surfaces can compress the 3-D image of a real object in a very efficient manner.