面向未知物体自动测量和重建的视点规划方法  

View Planning Method for Automatic Measurement and Reconstruction of Unknown Objects

作　　者：何炳蔚[1,2] 周小龙[1] 

机构地区：[1]福州大学机械工程及自动化学院,福州350108 [2]天津大学精密测试技术及仪器国家重点实验室,天津300072

出　　处：《计算机辅助设计与图形学学报》2010年第8期1374-1381,共8页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(50605007);天津大学精密测试技术及仪器国家重点实验室开放基金;福建省高校新世纪优秀人才支持计划(XSJRC2007-07)

摘　　要：针对未知三维物体自动测量和重建问题,提出一种基于视觉系统极限可视区域的新颖视点规划方法.首先确定单目线激光视觉测量系统的最佳可视测量区域,结合初始视点下所获模型的边界区域信息构造未知空间的极限可视表面模型,并以此表面作为未知物体的预测曲面模型;然后确定下一视点预测曲线的可视性,将左右规划进程中能获取最大可视曲线长度的位置定义为下一最优视点的参考位置,通过比较左右参考位置下所获可视面积的大小,取大者所在位置作为下一视点的最终位置.最后通过对实体模型重建,验证了文中方法的可行性及有效性.In order to the automatic measurement and reconstruction of the unknown three dimensional objects, a new view planning method based on the limit visible region of the vision system is proposed in this paper. Firstly, the visible and measured region of the monocular line-laser vision system is obtained, and the limit visible surface is modeled by the known boundary data obtained from initial view and is used to predict the surface information of the unknown object. Then the visibility criterion of next viewpoint is determined according to the predict curve, and the position that could obtain the maximum visible curve length in left or right planning process is defined as the next best view（NBV） candidate position. The larger one is selected as the final NBV position by comparing the visible area of the above two candidate positions. Finally, we have experimentally validated our research with some experiments for reconstruction.

关 键 词：三维重建 视点规划 极限面 下一最优视点 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]