摘要
利用人体行走的单目动态图像序列 ,估计手臂及腿部的运动及结构参数 .在物体中心坐标的运动表示方式下 ,提出了基于弹性连接概念的非刚体运动模型 ,通过加入弹性系数的方法 ,使非刚体运动模型和刚体运动模型得到了良好的统一 .在此模型的基础上 ,用 Levenberg- Marquardt方法实现了运动及结构参数的估计 .实验证明了该方法的有效性 。
An approach to the estimation of the motion and structure parameters of brachium and leg based on monocular image sequence is proposed. By using the motion description under the object-centered coordinate system, a non-rigid motion model based on the notion of elasticity is developed. The non-rigid motion model and the rigid one are well unified by incorporating the elasticity parameters into the rigid motion model. Based on this model, the motion and structure parameters are estimated using the method of Levenberg-Marquardt. Results show that the method is effective and the motion pattern can be reflected to some extent by the elasticity parameters.
出处
《软件学报》
EI
CSCD
北大核心
2001年第11期1732-1738,共7页
Journal of Software
关键词
非刚体运动
相对弹性模型
结构参数估计
单目人体图像序列
医学
Mathematical models
Medical imaging
Motion estimation
Parameter estimation
Probability density function