A high precision 3D object reconstruction method using a color coded grid and NURBS

A high-precision technique for reconstructing three-dimensional (3D) objects from its one-camera 2D views and using a color-coded structured light is described. The method does not require any a-priori knowledge of the absolute positioning and orientation of the camera, nor of a point in the scene. However a calibration process precedes the 3D reconstruction algorithm leading to a high-precision calculation of the position of the camera. The structured light, which is pseudo-randomly color-encoded, is used to determine unique points and lines on the surface of the object. The lines once extracted from a view of the object, are used for the determination of non-uniform rational Bezier splines (NURBS) curves to generate the surface of the object. The precision of the technique depends on the adaptation of the sampling rate of the structured light to the surface of the object. In a recent test, the precision of 7 tau (thousands of an inch) was achieved with a relatively smooth object. The paper describes only the 3D reconstruction process and not the integration of multiple views into one single 3D image of the object.