基于各点异性理论的基础矩阵鲁棒估计

Robust estimation on fundamental matrix based on heteroscedastic theory

针对基础矩阵常用算法对噪声过于敏感、抗干扰能力差等缺点,基于误差与变量相关(Errors-in-Variables,EIV)模型提出1种新的鲁棒性基础矩阵估计算法.该算法采用各点异性回归技术,建立EIV模型,依据数据矢量观测集合最优地估计EIV模型参数和数据矢量真值集合.实验结果表明,在存在较大噪声干扰的条件下,此算法仍能较为准确地估计基础矩阵,具有良好的鲁棒性和较快的运算速度.

With the disadvantages of high sensitivity to noise and poor anti-jamming capability in many common fundamental matrix algorithms, a new algorithm is proposed based on EIV（Errors-in-Variables） model. In the algorithm, the heteroscedastic regression technique is adopted, the EIV model is established, and the optimal parameters of EIV model and truth-value set of data vectors are estimated according to the observation set of data vectors. The experiments show that the algorithm can estimate fundamental matrix robustly and precisely even if there is great noise.