针对基于三视图的传统三维重建时特征点匹配后保留正确匹配点对数量过少、由图像噪声、拍摄照片时遮挡和震动等因素引起的匹配误差较大等问题,文章提出了一种自适应加权迭代算法,该算法通过给匹配点对加权进行迭代筛选来实现,可以稳定的获得数量较多、结果较准确的匹配点对,从而精确求解三视图空间约束矩阵三焦点张量.实验证明,文章提出自适应加权迭代算法相比传统的RANSAC估计算法可以获得更加精准稳定的三视图匹配点对等结果,进而通过该算法可以得到更加理想的重建结果.%To solve the problem that the number of accurate matching points based on traditional three view 3D reconstruction is too small to retain, larger errors of feature points matching which are caused by image noise exist, the situation of covering and shaking when taking pictures and so on, this paper proposes an adaptive weighted iterative algorithm, which is implemented by iterative screening of the weighted matching points, this algorithm can accurately solve more three-view matching points and further the trifocal tensor. Finally, the experimental results show that the robust adaptive weighted iterative algorithm can get more accurate and stable results of the three-view matching points comparing with traditional RANSAC algorithm and thus get an ideal three-dimensional reconstruction results.
展开▼
