基于ICP方法缺乏纹理物体三维曲面模型的匹配

3-D surface model registration of object lacking of texture based on ICP

对缺乏纹理物体建立三维模型,关键的难度在于缺乏特征点和点的坐标获得。本文采用ICP算法来对物体进行匹配,当特征点不足时,ICP算法是一个很适用的匹配方法,只需要对物体的给定点寻找最近点。为了计算目标点的坐标,利用投影器投射清晰稳定的纹理到缺乏纹理物体的表面,然后CCD相机对物体进行拍摄。通过线检测和空间前方交会解算出物体部分模型表面的空间点;并使用无需控制点的ICP方法对相互重叠的部分模型进行匹配,从而完成物体的三维重建。实验证明ICP方法对缺乏纹理物体的三维曲面匹配有效。

In order to construct a 3D model of object lacking of texture, the main difficulty for registration is the lacking of feature points and the obtaining of the points' coordinates.In this paper a method is proposed based on the Iterative Closed Point (ICP) algorithm, which requires only the procedure to find the closed point on a geometric entity to a given point.The ICP algorithm is a popular method for the registration when lacking of feature points.In order to compute the points' coordinates, the projector can provide the clear and stable texture on the surface of the object lacking of texture easily.The camera is used to take photos as the image data for the next processing.Using curve detection and space intersection, spatial points on the surface of the sheetmetal parts are obtained.Sub-models overlapping each other are registered by ICP, so the 3D reconstruction is finished.The feasibility of ICP is verified by the results of the experimentation.