This paper presents a novel scanning-path determination method to choose a next best view, using a combination of the shape-from-silhouette and convex-hull methods to ensure the accuracy and integrity of measured object surface data. A backlight system is used to illuminate the scene to acquire silhouettes from different viewpoints, from which an approximate 3-D surface model can be deduced. The scanning path is determined according to the convex hull of the approximate model so as to specify the movement of a three-axis motion table to achieve automatic measurement. With a mathematical model of a sensor equipped with a color CCD and a line-structured laser, 3-D color data of the object can be obtained integrally and accurately.
A completed system for color 3D object measurement is presented, which is composed of a line-structured laser sensor and a two-axes translation stage. As well as the mathematics model of 3D measurement based on line-structured laser method is founded and two groups of unknown-parameters are derived, namely the camera parameters and the light-plane parameters. The color 3D data (xi,yi,zi)-(Ri,Gi,Bi) is obtained by merging the color data (Ri,Gi,Bi) and 3D (xi,yi,zi) data according to the corresponding pixel coordinates (ui,vi) of the stripe. A kind of circle pattern is applied in the calibration on both groups of parameters, however the way to acquire the control points by use of the pattern is different in each calibration process. In the calculation of light-plane calibration, an unrestricted object function is constructed to ensure the orthogonal character and accuracy of the parameters. In order to verify the measurement accuracy of the calibrated system, the measurement on a standard jig is performed with the accuracy better than 0.1mm. The result shows the parameters calibration is reasonably effective and reliable. The system is developed for the antique digitization, game development, etc., and it will have a promising future.