基于双目视觉的球形物体运动轨迹重建

Reconstruction of Spherical Object Motion Trajectory Based on Binocular Vision

文章介绍了一种改进的球形目标运动轨迹拟合算法,并以高尔夫球为实例进行算法验证。针对三次样条插值曲线平滑性较弱问题,提出一种基于滤波降噪的改进插值算法,实验结果表明该算法对插值结果影响细微的情况下平滑了拟合曲线。实验首先检测出球心坐标,在双目序列图像之间根据坐标差异进行空间点的三维重建;再用一种基于Savitzky-Golay滤波的三次样条插值算法将高尔夫球在空间中的三维散点进行非线性拟合处理,最终完成运动轨迹的三维重建。通过模拟实验,与传统算法相比,文章提出的改进算法恢复运动物体三维信息的效果更加显著,且平均误差率在5%以内。

This paper introduces an improved algorithm for fitting the movement trajectory of spherical targets,and validates the algorithm with a golf ball as an example.Aiming at the problem of weak smoothness of cubic Spline interpolation curve,an improved interpolation algorithm based on filtering noise reduction is proposed.The experimental results show that the algorithm smoothes the fitting curve when the effect of the algorithm on the interpolation result is slight.Firstly,the spherical center coordinates are detected,and the 3D reconstruction of spatial points is performed between binocular sequence images according to the coordinate difference;Then a cubic Spline interpolation algorithm based on Savitzky Golay filtering is used to nonlinear fit the golf ball's 3D scattered points in space,and finally the 3D reconstruction of the motion trajectory is completed.Through simulation experiments,compared with the traditional algorithm,the improved algorithm proposed in this paper has a more significant effect in restoring the 3D information of moving objects,and the average error rate is less than 5%.