基本矩阵作为分析两视图对极几何的有力工具，在视觉领域中占用重要的地位。分析了传统鲁棒方法在基本矩阵的求解问题中存在的不足，引入了稳健回归分析中的LQS方法，并结合Bucket分割技术，提出一种鲁棒估计基本矩阵的新方法，克服了RANSAC方法和LMedS方法的缺陷。模拟数据和真实图像实验结果表明，本文方法具有更高的鲁棒性和精确度。

The fundamental matrix is an effective tool to analyze epipolar geometry and plays an important role in computer vision. This paper analyzes the shortcoming of traditional robust methods in estimating the fundamental matrix, and proposes a novel technique for estimating the fundamental matrix using LQS and bucketing technique in robust regression, which eliminates the drawbacks of RANSAC and LMedS. Experimental results on synthetic data and real images show that the proposed algorithm achieves high accuracy and robustness.