稳定估计F矩阵的双对极点约束算法

从基本矩阵参数的几何意义出发，给出了新的估计基础矩阵的约束条件，使无约束规划的目标函数由原来的4元8次降为4元6次，并用求解非线性方程的方法求出了双对极点位置，实现了F阵的稳定估计，有效地避免了传统的用非线性最小二乘法求解所存在的不足。实验结果表明，文中所提供的方法彻底克服了原双对极点约束方法运算速度慢、存在大量伪解和对极点的不稳定现象，实用性较强。

A Robust Biepipole Constraint Algorithm for Estimating Fundamental Matrix

A new geometrical constraint, based on the geometrical analyzing of synthetic method, is developed to estimate fundamental matrix (F-matrix). A nonlinear equation of degree 8 in 4 unknowns for unconstraint optimizing objective function is reduced to a nonlinear equation of degree 6 in 4 unknowns by utilizing the constraint. Two epipoles can be obtained by solving the nonlinear equation. Compared to the traditional Least-Square method, the method is more robust. Experiments show that using the algorithm much more improvement in the speed of calculation can be obtained. Furthermore, problems raised by using two-epipole constrained algorithm, such as existing of pseudo-solution and poor stability of epipoles, can be completely eliminated. Hence, better practicability has bee n achieved.