基于鲁棒估计灭点分层重建的研究

Research of hierarchical reconstruction based on robust vanishing point

灭点是分层重建过程的重要信息,其求解的准确程度直接关系到最后三维重建的效果。提出了一种基于Hough算法的直线聚类检测方法求取图像中的直线信息以及基于RANSAC的由直线信息估计灭点信息的改进算法,以提高估计灭点的鲁棒性。经试验证明,将所提出的算法应用到分层重建的系统中,在仅有两幅图像的情况下能准确地对目标模型进行重建。

Vanishing point is the important information for a hierarchical reconstruction,and its solution accuracy is directly related to the the final results of reconstruction.The paper first proposes an linear clustering detection algorithm based on Hough to get the line information.Then proposes an improved vanishing point estimation method based on RANSAC which improves the robustness of vanishing point estimation.The test verifies that the algorithm is applied to the system of hierarchical reconstruction from a pair of views,and get the desired results.