基于3D骨架和MCRF模型的行为识别  

Human activity recognition based on 3D skeletons and MCRF model

作　　者：刘皓[1] 郭立[2] 易波[1] 王冠中[1] 

机构地区：[1]中国科学技术大学物理系,安徽合肥230026 [2]中国科学技术大学电子科学与技术系,安徽合肥230027

出　　处：《中国科学技术大学学报》2014年第4期285-291,共7页JUSTC

基　　金：国家自然科学基金(61071173)资助

摘　　要：针对目前行为识别方法的不足,提出一种基于人体3D骨架和多CRF模型(MCRF)的行为识别方法.3D骨架数据量少且保留了行为关键信息的优点,并具有融合多特征和上下文信息的优势.为此,首先基于3D骨架将人体动作划分为全局运动、手臂运动和腿部运动,通过对动作序列进行多类特征提取,形成多类特征集;然后利用CRF模型对每一特征集建模,再融合所有的CRF模型,得到MCRF模型;最后利用MCRF模型进行行为识别.实验结果表明,该方法具有较高检测率.Considering the disadvantages of the traditional human activity recognition system,a human activity recognition system using an MCRF model and 3D skeletons was proposed.Its 3D skeleton data has less data and retains the key information,and the MCRF model has the advantage of being able to combine more features and utilizing adaptive contextual information.First,human activity was divided into global activity,arm activity,and leg activity.Several feature subsets were formed through more feature extraction.Then,CRF models were used on each feature subset to generate CRF units.Finally,all the CRF units were combined to produce the MCRF model which was utilized to recognize human activity.The experimental results indicate that the proposed method can improve detection accuracy.

关 键 词：行为识别 3D骨架 MCRF 特征提取 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]