一种高精度估计的基础矩阵的线性算法  

A Linear Algorithm with High Accuracy for Estimating Fundamental Matrix

作　　者：陈泽志[1] 吴成柯[1] 

机构地区：[1]西安电子科技大学综合业务网国家重点实验室,陕西西安710071

出　　处：《软件学报》2002年第4期840-845,共6页Journal of Software

基　　金：国家自然科学基金资助项目(69972039;60002007);香港研究基金委员会资助项目(CUHK4310/98E);中法先进研究计划项目(PRA SI 00-04)

摘　　要：通过引入与余差有关的代价函数,给出了一种高精度估计基础矩阵的线性算法——加权平移算法.首先将原始输入数据加权,计算加权后数据的重心坐标,将坐标原点平移到该重心坐标,再作归一化处理.然后用8点算法求出基础矩阵F阵的8个参数,实现了F阵的高精度估计.实验结果表明,此算法具有良好的鲁棒性,且余差和对极距离都小于其他线性算法,提高了基础矩阵的精度.In this paper, a cost function relative to residual errors is introduced, and a linear algorithm by exploiting the strategy of weighted translation transformation is presented. Firstly, the original input data is weighted and the centroid coordinates are calculated, and the origins of coordinates are translated to their centroids. Then, the matching points are normalized. Finally, the eight parameters of fundamental matrix (F-matrix) can be solved and the procedure of estimating the fundamental matrix with high accuracy can be achieved. Experimental results show that this algorithm performs very well in terms of robustness to outliers and noises. The algorithm is superior to other algorithms in residual errors and average epipolar distance and improves the accuracy of F-matrix.

关 键 词：对极几何 基础矩阵 加权平移 线性算法 鲁棒性 计算机图形学 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]