Abstract: A robust 3D reconstruction method based on image sequence was presented. First, an optimal parameter estimation of two images was applied. Then, along with the number of images increasing, the sparse bundle adjustment was adopted to reduce the minimum geometric error of the measurements of image coordinates. Finally, the 3D structure and camera parameters were processed with a global optimization method to improve the robustness of the reconstruction. The experiment results demonstrate that the proposed method can effectively improve the accuracy and robustness of reconstruction, and factually represent 3D model of objects.

Key words: 3D reconstruction, robust estimation, sparse bundle adjustment, image sequence, global optimization

摘要： 提出了基于序列图像的鲁棒三维重建方法。首先利用两幅图像的最优参数估计，然后添加新图像并采用稀疏调整，减少图像坐标测量值的最小几何误差。通过对三维结构和摄像机参数进行全局优化处理，以提高重建的鲁棒性。实验结果表明，该方法提高了重建的精度和鲁棒性，并真实地再现了物体的三维模型。

关键词: 三维重建, 鲁棒估计, 稀疏捆集调整, 序列图像, 全局优化