基于整体-星型模型的人体姿态估计  

Human Pose Estimation with Global-constellated Model

作　　者：代钦 石祥滨[3] 乔建忠 刘芳[2,3] DAI Qin;SHI Xiang-bin;QIAO Jian-zhong;LIU Fang(College of Information.Shenyang Institute of Engineering,Shenyang 110136,China;College of Computer,Northeastern University.Shenyang 110819.China;Department of Computer.Shenyang Aerospace University,Shenyang 110136,China)

机构地区：[1]沈阳工程学院信息学院,沈阳110136 [2]东北大学计算机学院,沈阳110819 [3]沈阳航空航天大学计算机学院,沈阳110136

出　　处：《小型微型计算机系统》2018年第12期2687-2693,共7页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(61170185)资助;辽宁省教育厅科学研究一般项目(L201602;LQN201709)资助

摘　　要：针对姿态估计方法受到人体姿态多样性的影响问题,本文研究适合表示人体结构的模型,提出一种基于多模整体模型和星型模型的人体姿态估计方法.首先,使用人体整体检测器完成人体整体方位的检测,以获得初始人体候选集;然后提出一种多模型的星型结构,表示具有一定方位偏差的人体各个部位和整体之间的关系;为了更好地描述相邻部位之间的分布关系,在整体-星型模型中融合树型模型.数据集IP和LSP的结果表明,该方法提高了姿态估计准确率,尤其提高在人体姿态多样的情况下的结果.A human body model suitable for expressing the body structure is studied aiming at the problem that the pose estimation is affected by the diversity of human poses. A human pose estimation method based on global multi-model and constellated multi-model is represented. Firstly,the initial candidate for global human orientation is obtained by global detector for human body. Secondly,constellated multi-model is proposed to depict the relationship between the human global orientation and diversified body parts. Thirdly,the global-constellated model and tree model is fused together in order to better describe the distribution relationship between adjacent parts. The experimental results on dataset IP and LSP show that the method improves the accuracy of pose estimation,especially for diverse pose.

关 键 词：姿态估计 星型模型 人体结构 可形变部位模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]