多视觉动画图像三维重建特征点匹配优化仿真  

Multi Visual Animation Image 3D Reconstruction Feature Point Matching Optimization Simulation

作　　者：李金明[1] 

机构地区：[1]西安外国语大学艺术学院,陕西西安710128

出　　处：《计算机仿真》2017年第9期341-344,共4页Computer Simulation

摘　　要：多视觉动画图像的三维重建,可有效解决复杂现实场景研究难度大的问题。对虚拟动画图像进行三维重建,需要对各视点的动画图像进行特征点提取,并对其进行匹配与校正,完成对动画三维图像的重建。传统方法获得图像每帧的三维点集,对三维点集进行融合处理,但忽略了对特征点的校正,导致重建效果不理想。提出基于稀疏光束平差理论的多视觉动画图像三维重建方法。利用张正友标定理论对多视觉动画图像进行标定,得到相机标定模块的内外参数,通过运用Harris角点检测算子对各视点的动画图像进行特征点提取,再利用最小平方中值法对提取的特征点进行匹配与校正,并完成多视觉动画图像三维重建。实验结果表明,所提方法可以快速精确地获得多视觉动画图像三维重建结果,且具有一定的实时性和可靠性。A three-dimensional （3D） reconstruction method for multi-vision animated image is presented baaed on adjustment theory of sparse light beam. The theory of Zhengyou Zhang calibration is used to calibrate the multi-vi- sion animated image, and interior and exterior parameters of calibration module of camera are obtained, then extrac- tion of feature point for animated image of each point of sight is carried out via Harris corner detection. Moreover, the median method with least square is used to match and correct the extracted feature point. Thus, the reconstruction is completed. Simulation results show that the method can obtain result of 3D reconstruction quickly and precisely. It has certain instantaneity and reliability.

关 键 词：多视觉动画图像 三维重建 特征点匹配 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]