基于图像特征的铅球投掷数据同步方法  

Data synchronous method for shot-put based on image feature

作　　者：任泓宇 聂余满[2] 王为田[2,3] 汪步云[2,3] 李世银[1] 宋全军[2] 

机构地区：[1]中国矿业大学信息与电气工程学院,江苏徐州221008 [2]中国科学院合肥智能机械研究所,安徽合肥230031 [3]中国科学技术大学自动化系,安徽合肥230026

出　　处：《计算机工程与设计》2014年第5期1699-1704,共6页Computer Engineering and Design

基　　金：国家体育总局重点研究领域课题基金项目(2012B080);国家自然科学基金项目(61201400;61105090)

摘　　要：为了在不干扰正常训练的情况下,实现数字铅球系统中的"力-姿"信息同步,提出了一种基于图像特征的非侵入式同步方法。采集高速图像序列后,利用圆度特征判决铅球出手帧,通过ROI设置及关键帧预判等方法提高算法效率,实验中对比分析了参数设置对算法效率的影响。实验结果表明,算法实现了高速高清图像的实时处理,同步准确率达到93.1%,有效地满足了辅助训练的需要。To solve the ＂force-posture＂ synchronization problem in the digital shot system without disturbing the normal training, a non-invasive data-image synchronization method based on image features is proposed. The key frame of shot-putting is judged by the roundness feature after the high-speed image sequences captured, and the approaches such as ROI setting and key frame predicting are employed to improve the algorithm efficiency for real-time high-speed and high-resolution processing. Then the influence of parameters setting on algorithmic efficiency is analyzed comparatively in experiments. The experimental results show the synchronization accuracy can reach 93.1 %, which meet the needs of auxiliary training effectively.

关 键 词：数字铅球 图像特征 “力-姿”同步 关键帧 圆形检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]