民族舞蹈运动数据的实例检索方法  

An Example-Based Method for Chinese Dance Motion Retrieval

作　　者：陈松乐[1] 孙正兴[1] 张岩[1] 李骞[1] 

机构地区：[1]南京大学计算机软件新技术国家重点实验室,南京210093

出　　处：《计算机辅助设计与图形学学报》2014年第2期198-210,216,共14页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(61272219;61100110;61321491);教育部新世纪优秀人才资助计划(NCET-04-0460);江苏省科技计划(BE2010072;BE2011058;BY2012190;BY2013072-04);计算机软件新技术国家重点实验室创新基金(ZZKT2013A12)

摘　　要：针对民族舞蹈运动的特点,提出一种基于姿态时空特征的民族舞蹈运动数据检索方法.首先采用姿态时空特征作为运动内容表示,有效地反映了舞蹈动作的局部性、节奏性以及细分性,为实现民族舞蹈运动数据检索提供了基础;其次采用无监督的多变量时间序列特征选择方法去除线性冗余特征,有效地提高了检索效率;最后利用基于权重调整的相关反馈机制捕捉用户的查询兴趣,改善检索性能.实验结果表明,该方法具有较好的检索性能,能够实现民族舞蹈运动数据检索.We present a method to perform Chinese dance motion retrieval based on pose space-time features according to the characteristics of Chinese dance movements. First, pose space-time features effectively reflect the locality, rhythm and subdivision of Chinese dance movements and provide a basis for performing Chinese dance motion retrieval. Second, an unsupervised feature selection algorithm for multivariate time series is introduced to remove redundant features and enhance the retrieval efficiency. Third, the relevance feedback based on feature weight adjustments is adopted to capture users＇ intention and improve the retrieval performance. Experimental results show that this method has better retrieval performance and can perform Chinese dance motion retrieval effectively.

关 键 词：民族舞蹈 运动数据 姿态时空特征 特征降维 相关反馈 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]