作　　者： ;

机构地区： 中山大学信息科学与技术学院无线电电子学系

出　　处： 《中山大学学报（自然科学版）》 1997年第3期48-53,共6页

摘　　要： 提出一种新的三维重建方法，对所获得的立体象对进行适当的变换，把空域中的立体匹配问题转换到波数域中来解决；采用基于局部相的立体匹配方法来求解立体匹配这个难题，并用Ｈｏｐｆｉｅｌｄ神经网络来实现立体匹配过程，以便实现并行处理； A new 3D reconstruction approach is presented. First, the stereo image pair obtained by vision system is transformed, and then the local phase is obtained for stereo matching processes. The Hopfield neural network is established, so that the computation can be implemented efficiently in parallel. A 3D reconstruction neural network is constructed by using BP neural network. The 2D points on images can be converted into 3D points on objects by using 3D reconstruction neural network. At last, the examples for both synthetic and real image are shown in the experiment, and good results are obtained.

关 键 词： 智能装配 计算机视觉 三维重建 视觉系统

领　　域： [金属学及工艺] [自动化与计算机技术] [自动化与计算机技术]