基于图像视频序列分析的乒乓球机器人击球轨迹预测方法  

A Method for Predicting the Hitting Trajectory of Table Tennis Robot Based on Image and Video Sequence Analysis

作　　者：张雪琴[1] 江帆[1] 任栋 ZHANG Xueqin;JIANG Fan;REN Dong(Xi’an Jiaotong University City College,Xi’an 710018,China;Xi’an Medical University,Xi’an 710021,China)

机构地区：[1]西安交通大学城市学院,西安710018 [2]西安医学院,西安710021

出　　处：《自动化与仪器仪表》2024年第8期99-103,共5页Automation & Instrumentation

基　　金：陕西省体育局常规课题项目研究关于体教融合背景下陕西省青少年体质健康路径和策略研究(2022017)。

摘　　要：为提高对乒乓球机器人击球轨迹的分析能力,提出并设计一种基于图像视频序列分析的乒乓球机器人击球轨迹预测方法。先对乒乓球机器人击球轨迹的图像视频序列进行采集,然后依据采集的图像视频序列进行乒乓球机器人击球轨迹动态点跟踪融合,得到其击球轨迹,最后构建乒乓球机器人击球轨迹的帧采样序列重构模型,通过空间视觉信息增强的方法,进行图像视频序列的结构参数检测,使用梯度下降法进行区域像素点检测,根据像素跟踪结果实现对机器人击球轨迹的预测。测试表明,采用该方法进行乒乓球机器人击球轨迹预测的精度为98.7%,预测耗时低于11.3 ms,优于对比方法,提高了对击球轨迹的动态预判能力。To improve the analysis ability of the hitting trajectory of table tennis robots,a method for predicting the hitting trajectory of table tennis robots based on image video sequence analysis is proposed and designed.Firstly,the image video sequence of the hitting trajectory of the table tennis robot is collected,and then the dynamic point tracking fusion of the hitting trajectory of the table tennis robot is performed based on the collected image video sequence to obtain its hitting trajectory.Finally,a frame sampling sequence reconstruction model of the hitting trajectory of the table tennis robot is constructed,and the structural parameters of the image video sequence are detected through spatial visual information enhancement,The Gradient descent is used to detect the regional pixel points,and the prediction of the robot hitting trajectory is realized according to the pixel tracking results.Tests have shown that the accuracy of using this method for predicting the hitting trajectory of table tennis robots is higher than 98.7%,and the prediction time is less than 11.3 ms,which is superior to the comparison method and improves the dynamic prediction ability of the hitting trajectory.

关 键 词：图像视频序列 乒乓球 机器人 击球轨迹 预测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]