基于序列模式挖掘的田径视频分类

A Novel Classification Based on Sequential Pattern Mining in Track and Field Videos

现有的体育视频分析方法大多都专注于重要事件的提取,而忽视了如何对这些事件进行组织和语义分析。本文提出了一种基于序列模式挖掘的田径视频镜头分类算法。本文主要围绕两个问题展开——特征提取和语义规则的定义。在特征提取阶段,自动的将田径视频镜头分割为一系列可识别的运动事件序列,然后使用机器学习的算法对每类行为事件进行识别。在语义规则定义阶段,使用序列模式挖掘方法发现其中的频繁序列,在此基础上进行。实验选用了上千段田径视频镜头进行测试,结果显示了本文算法进行田径视频镜头分类的有效性。

The majority of existing work on sports video analysis concentrates on highlight extraction.Little work focuses on the important issue as how the extracted highlights should be organized.In this paper,we present a shot classification approach based on sequential pattern mining to automatically segmenting a video into a sequence of recognizable action events.Two research challenges of highlight ranking are addressed,namely affective feature extraction and semantic analysis.A set of visual features,are extracted to recognize action event by machine learning.We mainly use effective frequent sequential pattern mining algorithm to dispose of the video classification and form a classification rule-lib to match the shot sequence that will be classified.More than 1000 video sequences are used to evaluate the performance of the proposed approach.The experimental results indicate that the good performances by the proposed method.