基于视频动作语义关联的视频复杂动作场景检测

Video Complex Action Scene Detection Based on Video Action Semantic Association

随着视频监控技术与互联网应用的普及,视频数据挖掘已成为当前大数据领域的重要研究内容之一。在视频数据挖掘中,对视频内容的理解需要从局部动作语义理解上升到更高层的复杂场景或事件语义理解。在对视频基本动作语义概念理解的基础上,提出一种基于基本动作语义概念关联的视频复杂动作场景检测方法。该方法首先记录视频动作语义概念的出现情形,对相应视频场景中的所有动作语义概念采用Aproiri算法进行关联规则挖掘,然后利用挖掘得到的动作语义概念关联规则,定义视频复杂动作场景检测分类准则,最后对测试视频采用该分类准则进行视频复杂动作场景概念检测。通过在典型数据集上的实验结果表明,该方法可以有效挖掘出视频中动作之间的关联关系,并实现对视频复杂动作场景概念的检测分类。

With the popularity of video surveillance and Internet applications,video data mining has become one of the current big data contents.The semantics of complex action scenes from the local action semantics to the higher level is needed to understand videos.On the basis of the understanding of the semantic concepts of the basic motions in videos,this paper proposes a video complex action scene detection method based on the concept association of basic action semantics.This method records the appearance of the concept of video action semantics,uses Aproiri algorithm to mining all the semantic concepts in the corresponding video scene,and defines the association rules of the concept of action semantics,defines the classification criteria of video complex action scene detection,and uses this classification criterion to detect the concept of complex action scene in video.Experiments are carried out on a typical dataset,and the experimental results show that the proposed method can effectively conduct mining of the correlation between actions in video and realize the concept detection and classification of video complex action scenes.