体育训练过程中的运动视频分析与识别研究

发布时间：2018-04-15 10:18

  本文选题：体育训练 + 运动视频　； 参考：《现代电子技术》2017年11期

【摘要】：当前的体育训练过程中,训练员使用难度较小的视频重播与解析管理方式为运动员讲解动作要领,不够直观和科学,不能满足训练员对运动效果评估的需求。针对该问题,研究了体育训练过程中的运动视频分析与识别过程,采用基于粒子滤波预测的自适应阈值运动目标分离算法实现运动目标的自适应分离。通过粒子滤波技术跟踪运动员的运动,塑造运动模型,并依据运动模型预测后续运动视频帧内不同重要关节点的位置,完成后续运动视频帧的跟踪。采用条件随机场方法实现体育训练视频中的动作识别。实验结果说明该方法具有较高的动作识别率和较低的误分离率。
[Abstract]:In the current sports training process, the trainers use the less difficult video replay and analysis management to explain the main points of action for the athletes, which is not intuitive and scientific enough to meet the needs of the trainers for the evaluation of sports effects.In order to solve this problem, the process of motion video analysis and recognition in the process of sports training is studied, and the adaptive threshold motion target separation algorithm based on particle filter prediction is used to realize the adaptive separation of moving objects.The particle filter technique is used to track the movement of athletes, to shape the motion model, and to predict the position of different key points in the frame of the subsequent motion video according to the motion model, so as to complete the tracking of the following motion video frame.The method of conditional random field is used to realize motion recognition in sports training video.The experimental results show that the method has higher recognition rate and lower error separation rate.
【作者单位】： 黄河科技学院;
【基金】：河南省科技厅科技攻关项目(132102310462)
【分类号】：TN713;TP391.41
，

本文编号：1753696