体育赛事数字视频直播的镜头转换检测算法分析

Analysis of shot conversion fetection algorithm for digital video live broadcast of sports events

文章基于体育赛事,对其数字视频直播的镜头转换的检测算法进行了分析。通过对累积帧间差异是否发生镜头转换进行判定,基于累积帧进行算法改进,对渐变与切变镜头做出的判断更加有效。在提取帧图像特征时,用均值差对单点像素值进行代替,本文算法可全部检测到突变,7个误检镜头,3个渐变漏检。在切变检测上,本文算法的查全率一定的提升,查全率达到94. 29%,查准率达到82. 86%,本文算法检测结果要比原算法更好。

Based on sports events,this paper analyzes the detection algorithm of digital video live shot conversion.It is more effective to judge whether the cumulative frame difference occurs or not,and the algorithm is improved based on the accumulated frame,and it is more effective for the judgment of the gradient and the cut lens.When the frame image features are extracted,the mean difference is used to replace the single point pixel value.This algorithm can detect all the mutations,7 false detection lenses and 3 gradual leakage detection.In the shear detection,the recall rate of the algorithm is improved,the recall rate is 94.29%,and the precision reaches 82.86%.This algorithm is better than the original algorithm.