图像重构中估计基础矩阵的线性迭代算法

A Linear Iterative Algorithm For Estimating Fundamental Matrix In The Imagebased Reconstruction

基础矩阵给出了图像间对极几何关系的代数描述,对其精确估计一直是未标定图像序列重构中的一个非常关键的问题.本文在8点算法的基础上,通过引入与对极距离有关的权因子,给出了一种高精度估计基础矩阵的线性迭代算法.实验结果表明,与归一化8点算法相比,此算法具有更强的抗噪声能力,更小的计算误差,从而提高了所估计基础矩阵的精度.

It＇ s a very important problem to estimate the fundamental matrix in the 3D reconstruction based on the un-calibrated image sequences. The fundamental matrix has a very good algebra description for the epipolar geometry relationship between two perspective images of single scene . In this paper, a weighted function relative to the epipolar distance is introduced, and a linear iterative algorithm with high accuracy based on the 8-point method is presented, Experimental results show that this algorithm performs very well in terms of robustness to noises. The algorithm is superior to the normalized 8-point algorithm in the residual errors and improves the accuracy of the fundamental matrix.