节点文献

基于线激光—单目视觉系统可视空间的未知物体三维重建方法研究

Research on Reconstruction Method for Unknown3-D Objects Based on the Visual Space of the Line-laser Monocular Vision System

【作者】 周小龙;

【导师】 何炳蔚;

【作者基本信息】 福州大学 , 测试计量技术及仪器, 2010, 硕士

【摘要】 随着先进测量设备和新型测量技术的发展,逆向工程技术在航空、航天、造船、汽车和模具等领域中的应用受到越来越多的重视和研究。而在逆向工程技术中,针对未知三维物体的自动测量和重建技术是研究的难点和热点,因此本文提出一种基于单目视觉系统可视空间的视觉规划新方法,使视觉系统能够根据物体表面部分已知信息,正确、高效地完成对未知三维物体的自动测量和重建。首先,通过实验获取所用线激光—单目视觉系统的可视空间,并在此基础上构建出系统的极限可视空间模型;其次,提出视觉系统可视空间和极限面相结合的视点规划策略。以极限面来预测未知模型的最大表面延拓信息,并通过确定下一视点可视性判据来获取下一视点可视的位置参数信息(旋转角度和平移量),从而确定能获取下一视点最大可视区域投影面积的参考位置,最终将能获取最大可视投影面积的位置定义为传感器在空间的下一最优视点位置;再次,为了保证规划过程的自终止性,提出两步规划自终止策略,即角度判据和表面数据完备判据,利用角度判据来终止对物体表面的初步重建,并利用点间距来判断已获物体表面是否存在遮挡信息,同时利用相邻点心距的差值作为确定遮挡规划方案的依据,从而获取遮挡部分信息,完成对未知三维物体的自动重建;最后,通过对CAD测试模型和多个实体模型的自动重构,验证本文方法的可行性和有效性。

【Abstract】 With the development of advanced measuring equipment and novel measuringtechnology, the reverse engineering has been of great importance applications inthe areas of aviation, space, shipbuilding, automobile, mould and so on. It hasattracted more and more attentions and researches. While in the reserveengineering, the automatic measurement and reconstruction for unknownthree-dimensional objects is a focusing topic. So this paper proposed a new viewplanning method based on the visual space of the monocular vision system. Thevision system can correctly and efficiently fulfill the automatic measurement andreconstruction of unknown3-D objects according to the partly knowninformation of the objects.Firstly, the visual space of the adopted line-laser monocular vision system isobtained. Then the limit visual space of the system is modeled.Secondly, a new view planning strategy that incorporates the visual space ofthe vision system with the limit visual surfaces is proposed. The maximal surfaceextension information of the unknown model is predicted by the limit visualsurface. And the position parameters (rotation angle and translation distance)that ensure the visibility of next view are obtained by acquiring the nextviewpoint visibility criterion. Then the candidate positions that could obtain themaximal projected area of the visual region at next view are determined. Thenext best view of the sensor is the position that can obtain the maximal projectedarea of the visual region among the candidate viewpoints.Thirdly, to terminate the planning process automatically, a two-step planningself-termination criterion is presented. Namely, angle termination criterion andsurface data integrity criterion are presented. The initial reconstruction is terminated by using the angle termination criterion. And the occlusion parts ofthe object are determined by the point pitch. Simultaneously, to obtain theocclusion information, the occlusion planning method is selected according tothe distance that from the point to the rotary-table center. Then thereconstruction process for unknown3-D object is completed automatically.Finally, the automatic reconstructions for the CAD testing model and theactual objects are experimented to validate the feasibility and efficiency of theproposed method in practical implementation.

  • 【网络出版投稿人】 福州大学
  • 【网络出版年期】2015年 06期
节点文献中: 

本文链接的文献网络图示:

本文的引文网络