基于单目视觉可视空间的视点规划方法  

View planning method based on visual region of monocular vision system

作　　者：何炳蔚[1,2] 周小龙[1] 

机构地区：[1]福州大学,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：《仪器仪表学报》2010年第2期390-396,共7页Chinese Journal of Scientific Instrument

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;福建省高等学校新世纪优秀人才支持计划(XSJRC2007-07)资助项目

摘　　要：通过确定单目线激光视觉测量系统的可视区域,提出将系统可视区域与未知模型极限面相结合的未知三维模型自动重建新方法。依据初始视点下所获模型信息,构造未知空间的极限面模型,结合系统可视区域确定下一视点的可视性区间,并将可视性区间内能获取最大极限面面积的位置定为下一个最优视点的位置。最后,通过对实体模型的自动重构,验证方法的可行性及有效性。Automatic reconstruction of unknown 3-D object is of great importance in areas of machine vision and visualization.In this paper,a new view planning approach of generating 3-D model is automatically proposed.The new algorithm incorporates the visual region of laser-vision system with the limit surface of unknown model and selects the suitability of viewpoints as the NBV on scanning coverage.The limit surface is modeled by means of the known data obtained from initial view.Then the visibility zone of the next view is determined.And the position that can obtain maximum limit surface area in the visibility zone is defined as the next best view position.Experimental results show that the method is effective in practical implementation.

关 键 词：视点规划 传感器规划 极限面 3D重建 下一最优视点 

分 类 号：TP391.72[自动化与计算机技术—计算机应用技术]