论文部分内容阅读
为了改进Kinect获取骨骼坐标的稳定性,提出了一种基于二阶指数平滑滤波的方法,引入移动速度趋势增量,并根据当前速度情况,采用一次线性插值方法实现平滑系数的自适应化,最后通过关节角度计算来获取人体上半身的关键部位角度,并映射到仿人机器人本体。实验结果表明,该算法拥有更好的平滑过渡性,并且能够有效地避免由于原始数据突变带来的仿人机器人手臂抖动的情况。
In order to improve the stability of Kinect for acquiring skeleton coordinates, a method based on the second-order exponential smoothing filter is proposed, which introduces the incremental trend of moving speed and uses a linear interpolation method to adaptively smooth the coefficients according to the current speed. Finally, Through the calculation of the joint angle to obtain the angle of the human body’s upper part of the key parts, and mapped to the humanoid robot body. The experimental results show that the proposed algorithm has better smooth transition and can effectively avoid the humanoid robot arm shaking caused by the sudden change of the original data.