At the present time, a great number of applications exist that need the reconstruction of three-dimensional surfaces, such as: medicine, reverse engineer, industrial tasks and virtual world recreation. A system of reconstruction of surfaces was made up of three parts: capture of image, calibration of the camera and generation of a mesh from the cloud of three-dimensional points (3D). The data 3D is captured using the principle of triangulation between a camera or receiver, a generator of structured light (emitting laser) and the object. The capture of the image is made through a digital camera, having used a set of algorithms developed in Borland C++ Builder5. For the camera's calibration the method of flexible calibration was implemented (Zhang, 2002), which consists of a solution of closed form, followed by a nonlinear refinement based on the criterion of maximal probability. In order to generate the three-dimensional surface a rectangular mesh was create using a set of B-spline curves.