中文摘要：

针对现有的三维运动估计算法在精度、效率和稳定性等综合性能上的不足,提出了一种结合双目视觉三维重建和利用对偶四元数表达运动参数的新算法。该算法以双目视觉系统为基础,采用SIFT算法进行图像特征点的提取和匹配;根据匹配关系进行三维特征点重建,以获取三维场景中运动目标的结构参数;利用对偶四元数可同时表示刚体的旋转和平移运动的特点,实现目标对象运动参数的表达和求解。通过实验将提出的算法与现有算法（包括奇异值分解法、正交分解法和单位四元数分解法）进行比较,结果表明,该算法具有更加简洁的表达形式,在保持传统算法精度和稳定性优势的基础上提高了计算效率,具有更优的综合性能。

英文摘要：

Considering the multiple requirements on accuracy, efficiency and stability of 3D motion estimation, a new algorithm combining 3D reconstruction of binocular vision and dual quaternion parameter expression is proposed. Basing on stereo vision system, the proposed algorithm firstly uses SIFT method to extract and match featuring points of images. Secondly, the 3D points are reconstructed to achieve the structure parameter of the moving target in 3D scene. Finally, dual quaternion, which can represent rigid translation and rotation simultaneously, is used to accomplish motion pa- rameter expression and calculation. Through contrastive experiments with the current methods including singular value decomposition, orthogonal decomposition and unit quaternion decomposition, the proposed algorithm with a more compact modus, performs better overall results of efficiency improvement, which maintains the accuracy and the stability at the same time.