3D regularized B-spline surface reconstruction from occluding contours of a sequence of images

摘要

The three dimensional surface reconstruction of a non polyhedral object is auddifficult problem in computer vision . In this paper, a new method for reconstructingudthree dimensional surface from the recovered motion of occluding contours isudpresented through calibrated image sequences . We use the uniform bicubic Bsplineudsurface patches to give a parametric representation of an object surface .udFinally, the problem of three dimensional B-spline surface patches reconstructionudis equivalent to find their control points by solving a nonlinear system . Twoudnumerical methods are outlined : Levenberg-Marquardt, Quasi-Newton . To avoidudthe classic camera calibration that needs a calibration pattern, we propose a directudnonlinear method of the autocalibration of a camera using the stable points in theudscene. Our approach can be applied in the case where the camera is calibrated,udthe object is smooth, specifically, that its surface is at least C2. Results forudreconstruction based on synthetic and real data are presented .