摘要
针对三维人体模型结构复杂,处理数量大且不易提取控制点等问题,提出通过对人体形状进行特征分析描述人体结构并进行姿态识别的算法。融合测地线与空间结构等特征提取骨架点有效减少数据的计算量,并通过ICP算法进行姿态的行为识别。实验证明,该算法有效地提升了三维姿态的识别效率,并有很好的鲁棒性。
Aiming at several problems such as the complex structure,large processing quantity and difficult extraction of control points of three-dimensional human body models,an algorithm of describing body structure and doing pose recognition by analyzing characteristics of shape of human body is proposed. It fused the characteristics of geodesic and spatial structure to extract skeleton points which reduce the calculation effectively and make behavior recognition of attitude by using ICP algorithm. Experiments proved that using this algorithm could recognize the three-dimensional pose more effectively with good robustness.
出处
《计算机应用与软件》
2017年第1期170-174,共5页
Computer Applications and Software
基金
辽宁省高等学校优秀人才支持项目(LJQ 2013110)
关键词
三维人体
语义骨架
姿态识别
Three-dimensional human body
Semantic skeleton
Pose recognition