【摘要】：無序多視影像的三維重建對噪聲非常敏感,錯誤的匹配關系會影響重建的精度,甚至直接導致重建失敗。提出了一種穩(wěn)健的批處理重建算法,首先利用回路閉合約束剔除可能存在誤匹配的三視匹配元,然后以三視匹配元中的三焦張量約束代替?zhèn)鹘y(tǒng)算法的核線約束來計算所有影像旋轉矩陣和相機中心位置的全局最優(yōu)解。重建過程中引入高效的并查集算法來提取多視匹配點,并利用迭代線性三角形算法計算空間點的三維坐標。實驗結果表明,所提算法在重建效率和計算精度方面都能取得較好的結果。
[Abstract]:The 3D reconstruction of disordered multi-view images is very sensitive to noise, and the mismatch relationship will affect the accuracy of reconstruction and even lead to the failure of reconstruction directly. In this paper, a robust batch reconstruction algorithm is proposed. Firstly, the loop closure constraint is used to eliminate the tri-view matching elements which may have mismatch. Then the three-focal Zhang Liang constraint in the tri-view matching element is used instead of the kernel constraint of the traditional algorithm to calculate the global optimal solution of the rotation matrix and the camera center position of all images. In the reconstruction process, an efficient parallel search set algorithm is introduced to extract the multi-view matching points, and the iterative linear triangle algorithm is used to calculate the three-dimensional coordinates of the spatial points. The experimental results show that the proposed algorithm can achieve good results in terms of reconstruction efficiency and computational accuracy.
【作者單位】： 信息工程大學地理空間信息學院;
【基金】：國家973計劃(2012CB720000)~~
【分類號】：TP391.41
，

本文編號：2220506