基于关节结构依赖的三维人体姿态估计与优化策略  

3D human pose estimation and refinement based on joint structural dependencies

作　　者：汪彬姿 宁欣 疏洋 丁友东 Wang Binzi;Ning Xin;Shu Yang;Ding Youdong(Shanghai Film Academy,Shanghai University,Shanghai 200072,China;Shanghai Engineering Research Center of Motion Picture Special Effects,Shanghai 200072,China)

机构地区：[1]上海大学上海电影学院,上海200072 [2]上海电影特效工程技术研究中心,上海200072

出　　处：《计算机应用研究》2025年第4期1268-1273,共6页Application Research of Computers

基　　金：教育部产学合作协同育人资助项目(220504276232753,230801284291638)。

摘　　要：由于面临着固有的深度模糊和潜在的遮挡,单目三维人体姿态估计仍充满着挑战。为了缓解不确定性,提出了一个轻量高效的三维人体姿态估计模型,充分利用人体关节的结构依赖关系,通过两个局部约束增强策略对高自由度关节施加额外约束;设计了一个基于扩散模型的后处理框架,通过对多假设结果逐步去噪并加以骨长约束,优化带噪声的三维姿态。该方法在Human3.6M数据集上相比同类方法,在两个常用指标上分别提升了3%和4.5%,验证了其优越性。Facing inherent depth ambiguity and potential occlusions,monocular 3D human pose estimation remains challen-ging.To mitigate uncertainty,this paper proposed a lightweight and efficient 3D human pose estimation model that leveraged the structural dependencies of body joints and applied additional constraints to high-degree-of-freedom joints through two local constraint enhancement strategies.The method used a diffusion model-based post-processing framework to progressively denoise multiple hypothesis results and apply bone length constraints,optimizing noisy 3D poses.On the Human3.6M dataset,the proposed method outperforms similar approaches by 3%and 4.5%on two common metrics,demonstrating its superiority.

关 键 词：扩散模型 三维人体姿态估计 图卷积神经网络 TRANSFORMER 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]