基于未标定图像的三维重建技术研究-计算机科学与技术专业论文.docxVIP

摘 摘 要 从运动恢复三维形状是计算机视觉领域的中心问题。而非定标图象序列的运 动结构问题是近十年的研究热点。在理论方面，从多视角几何关系、射影重建到 相机自定标技术都得到长足进展。在应用方面，随着计算机及网络技术的飞速发 展，许多应用领域对三维图形数据提出更高的要求。如何尽可能自动地获取复杂、 逼真的三维图形模型是当前的一个前沿课题。本文针对这一课题进行探讨，在一 些非定标图像三维重建的关键算法上提出一些新的改进。其有效性在仿真数据和 真实场景图象序列实验中得到了证明。 首先，在角点检测方面，针对susAN角点算法在实际应用中的不足，提出了 改进的SuSAN角点算法，使角点均匀的分布在图像上，有利于角点的匹配。 其次，根据基本矩阵的计算往往转化为最优化问题，研究了基于遗传算法的 基本矩阵估计方法，对基本矩阵的鲁棒求解进行了一次有意义的尝试。 最后，通过摄像机白标定，利用矩阵分解的方法和三角形法建立了场景的三 维模型。通过实际图像的仿真实验，表明了本文方法的有效性。 关键词：susAN算法，基本矩阵，摄像机自定标，三维重建 AbstractThe Abstract The Structure fr伽Motion (SFM) problem is the central problem in computer vision c伽nunity．The uncalibrated structure from motion problem (USFM)is the hot topic in the recent lO years．Many theories， including 叫ltiple view geo嘴try， projective reconstruction and c锄era self—cal ibration etc． have been proposed and宵e1 1 developed． In the applications， as the techn0109ies of computer and Internet deVeloping， many applications require 3D graphic data of high quality．One ch811enging problem is how to obtain complicated and realistic 3D graphic models as auto衄tically as possible． This thesis is focused on this problem and improves some key algorithms in the USFM problem． The effectiveness of our algorithms is evaluated in the experiments with si叫lated data and real image sequences． Firstly，on the research of interest point detection，since霄hen use SUSAN algorit|1nl to detect real imge，the number of features is too舢ch， so we propose a improved SUSAN algorit hⅡI to decrease the number of corners and to control the density of features iH the iIIlage． Experiment proves our algorithm can reserve good features and match th鲫profitably： Secondly， in the computation of fund硼ental髓trix， it always turns to optimization problem． An algorithm based on genetic algorithm is researched and the attempt to get robust solution to fundaInental matrix makes sense： Finally。 after determine the camera’8 intrinsic c8libration paraIIleters． a Euclidean scene reconstruction is set up using嬲trix decomposition m