计算机视觉中基本矩阵的估计方法

基本矩阵的估计是计算机视觉中一项重要的研究课题.从特征点的提取、匹配到从一组对应点求出基本矩阵的估计值,完整地介绍了求解基本矩阵的方法.对于噪声干扰以及问题固有的复杂性所导致的误匹配,以及由此导致的估计结果恶化问题,采用最小中值法（LMedS）,有效地解决了存在误匹配情况下稳定、准确地估计基本矩阵的问题.

The estimation of the fundamental matrix in computer vision

The estimation of the fundamental matrix is a crucial problem in computer vision. A method is developed to estimate the fundamental matrix accurately, including extracting feature points, matching feature points and estimating the fundamental matrix from the matched feature points. Because the outliers by a false match can severely disturb estimation of the fundamental matrix, this paper introduces a robust algorithm based on least-median-squares. Results are presented for real image pairs, which shows that the algorithm performs very well in terms of robustness to outliers, so that excellent results can be obtained.