This paper proposes a new robust algorithm(PSO-LMedS) for estimating fundamental matrix.This new algorithm views matching points obtained by LMedS as the initial population in PSO,then improves fundamental matrix estimation accuracy by using global optimization characteristics of PSO.Finally,it uses the new approachestimate the fundamental matrix,and the simulation results show high robustness and accuracy.%该文提出了一种新的估计基础矩阵的鲁棒算法——基于粒子群算法的最小平方中值法(PSO-LMedS)。该方法是将最小平方中值法(LMedS)与粒子群算法(PSO)相结合，将LMedS方法得到的匹配点对作为PSO的初始种群，利用粒子群算法的全局优化特性，通过不断的寻优过程，估计得到最佳基础矩阵，提高基础矩阵估计精度。通过仿真实例，证明改进后的算法具有较高的鲁棒性和精确性。
展开▼
