摘要
以网球视频裁判系统为个案,针对体育赛事智能视频系统提出一种新型开发思路。使用深度卷积神经网络算法,对赛场视频系统获得的图像信息进行深度卷积处理,全新搭建赛场三维动态模型,而非直接利用视频数据的倾斜摄影建模技术进行体育赛事的赛场动态建模。仿真分析中,该系统将传统系统0.43%的判罚失误率压缩到0.11%,裁判员对传统系统的主观评价为7.862分,低于该系统的8.404分。最终认为该系统开发模式具有一定的推广价值。
Taking the tennis video referee system as a case, this paper puts forward a new development idea for the intelligent video system of sports events. The deep convolution neural network algorithm is used to convolute the image information obtained from the video system of the sports field, and a new three-dimensional dynamic model of the sports field is built, instead of the oblique photography modeling technology of the video data. In the simulation analysis, the system reduces the penalty error rate of traditional system from 0.43% to 0.11%, and the referee’s subjective evaluation of traditional system is 7.862, which is also lower than 8.404 of the system. Finally, it is considered that the system development mode has a certain promotion value.
作者
马丽
刘晓磊
MA Li;LIU Xiaolei(Bayin Guoleng Vocational and Technical College,Korla 841000,China)
出处
《微型电脑应用》
2023年第2期150-153,共4页
Microcomputer Applications
关键词
体育赛事
深度卷积神经网络
视频裁判系统
赛场建模
sports events
deep convolution neural network
video referee system
field modeling
