基于三维图像的运动员起跑动作误差预测仿真
发布时间：2018-07-15 16:29
【摘要】：田径比赛运动支撑期间,起跑动作的准确性可使运动员在最短时间内获得最佳速度,直接影响比赛成绩,因此需要对运动员起跑动作进行误差预测,当前方法预测起跑动作时,难以对运动员起跑动作关节点进行准确跟踪,降低了运动轨迹预测的精度。提出一种基于三维图像的运动员起跑动作误差预测方法。上述方法选用星型骨架组成结构描述运动员起跑动作三维运动模式,采用ISOMAP非线性降维方法计算获取起跑动作图像三维子空间的运动状态投影,将起跑动作三维数据投影至非线性低维子空间中,识别出运动员起跑动作状态的内在结构后研究整个起跑动作的各个关节点,利用Mean-Shift搜索算法确定运动员起跑动作各个关节点位置,并通过卡尔曼滤波器算法进行运动员起跑动作误差预测,确定运动员起跑过程中的动作轨迹。仿真结果表明,所提方法可有效提升运动员起跑动作轨迹预测精度,且预测效率较高。
[Abstract]:During the support of track and field sports, the accuracy of the starting movement can make the athletes get the best speed in the shortest time and directly affect the performance of the competition. Therefore, it is necessary to predict the starting movements of the athletes. The current method is difficult to accurately track the joint points of the starting movement and reduce the track of the movement. The accuracy of track prediction is proposed. A method of predicting the starting movement error of athletes based on three-dimensional images is proposed. The above method uses the star frame structure to describe the three dimensional motion pattern of the starting movement of the athletes, and uses the ISOMAP nonlinear dimensionality reduction method to calculate the motion state projection of the starting movement image in the three-dimensional subspace, and the starting movement is three The dimension data is projected into the nonlinear low dimension subspace. After identifying the inner structure of the starting movement state of the athletes, it studies the joint points of the whole starting movement, and uses the Mean-Shift search algorithm to determine the position of each joint point of the starting movement of the athletes, and the Calman filter algorithm is used to predict the starting movement error of the athletes. The simulation results show that the proposed method can effectively improve the prediction accuracy of athletes' start motion trajectory, and the prediction efficiency is higher.
【作者单位】： 西藏民族大学体育学院;
【分类号】：G822;TP391.41
，
本文编号：2124691
[Abstract]:During the support of track and field sports, the accuracy of the starting movement can make the athletes get the best speed in the shortest time and directly affect the performance of the competition. Therefore, it is necessary to predict the starting movements of the athletes. The current method is difficult to accurately track the joint points of the starting movement and reduce the track of the movement. The accuracy of track prediction is proposed. A method of predicting the starting movement error of athletes based on three-dimensional images is proposed. The above method uses the star frame structure to describe the three dimensional motion pattern of the starting movement of the athletes, and uses the ISOMAP nonlinear dimensionality reduction method to calculate the motion state projection of the starting movement image in the three-dimensional subspace, and the starting movement is three The dimension data is projected into the nonlinear low dimension subspace. After identifying the inner structure of the starting movement state of the athletes, it studies the joint points of the whole starting movement, and uses the Mean-Shift search algorithm to determine the position of each joint point of the starting movement of the athletes, and the Calman filter algorithm is used to predict the starting movement error of the athletes. The simulation results show that the proposed method can effectively improve the prediction accuracy of athletes' start motion trajectory, and the prediction efficiency is higher.
【作者单位】： 西藏民族大学体育学院;
【分类号】：G822;TP391.41
，
本文编号：2124691
本文链接：https://www.wllwen.com/jiaoyulunwen/tylw/2124691.html
