一种估计运动刚体旋转参数的鲁棒方法

A New Robust Algorithm lor Estimation of 3-D Rotation Parameters ol Rigid Objects

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 motion analysis of computer vision, estimation of 3-D rotation parameters is a very important subject. Unfortunately, the existing algorithms for this subject have some disadvantages, especially in robustness. In this paper, a new robust algorithm has been proposed, based on the idea of making full use of constraints or priori information on the problem. The algorithm only involves linear operations instead of solving nonlinear equa- tions. Moreover, the algorithm has advantages than the existing algorithms in numerical accuracy, robustness, and orthogonality of rotation matrix. The results are very satisfactory with noisy synthetic data.