高动态舞蹈视频中特定动作片段检索系统设计

Design of retrieval system for specific action fragments in high-dynamic dance video

为了实现高动态舞蹈视频的分析和管理,需要对视频中特定动作片段检索系统设计方法进行研究。当前基于镜头内容的高动态舞蹈视频特定动作片段检索系统设计方法存在检索效率较低的问题。为了减少检索系统运行时间,提高查准率,提出一种基于相似性计算的高动态舞蹈视频特定动作片段检索系统设计方法。首先对高动态舞蹈视频中特定动作片段的背景进行初始化处理和更新,获取特定动作片段前景图像,实现其特征检测;然后依据特征检测结果,计算特定动作片段的密度分布、紧密程度和离散程度,以及特定动作片段的活跃块比率和它们之间的相似性;最后在此基础上设计特定动作片段的检索系统框架。实验结果分析证明,所提方法能够提高查全率和查准率,且减少了检索系统运行时间。

In order to analyze and manage the high-dynamic dance video,it is necessary to study the design method of retrieval system for specific action fragments in video. The current lens content based design method of specific action fragments retrieval system for high-dynamic dance video has the problem of low retrieval efficiency. In order to shorten the running time of the retrieval system and improve the precision ratio,a similarity calculation based design method of specific action fragments retrieval system for high-dynamic dance video is put forward. The initialization processing and update were performed for the specific action fragments background in high-dynamic dance video to acquire the foreground image of the specific action fragments,and realize its feature detection. According to the results of the feature detection,the density distribution,close degree and discrete degree of the specific action fragments,as well as active block ratio and similarity of the specific action fragments are calculated. On this basis,the framework of the specific action fragments retrieval system was designed. The experimental results show that the proposed method can improve the recall ratio and precision ratio,and shorten the running time of the retrieval system.