在计算机视觉领域中,估计运动物体的旋转参数是其中的重要内容之一。本文提出了一种新的估计旋转参数的鲁棒方法,这种方法避免了求解非线性方程组,所采用的都是线性运算,不仅如此,这种方法在数值准确性、鲁棒性和保证旋转矩阵的正交性方面都优于现存的方法。用模拟数据进行了大噪声条件下的实验,结果是令人满意的。 In the field of computer vision, estimating the rotation parameters of moving objects is one of the important contents. In this paper, a new robust method for estimating rotation parameters is proposed. This method avoids solving linear systems of equations by using linear operations. Not only that, but also the numerical accuracy, robustness and guarantee The orthogonality of the rotation matrix is ​​superior to the existing methods. Experiments under high noise conditions were performed with simulated data and the results were satisfactory.