摘要
由于非刚体运动形变估计是一个非线性的逆问题,非刚体运动时受到背景多样性、运动复杂性、数据的丢失、特征点噪声等问题的影响,使非刚性物体的形变程度估计面临着估计不准,误差较大的问题。为了解决这些问题,提出一种基于特征运动簇反投影的非刚体运动形变估计算法,将非刚体运动的形变估计过程中,加入一定的物理运动约束刻画不同非刚体运动形状基的情况下,运动形变的程度,使用非线性优化闭合求解的方法进行运动模型结构参数的重构,运用反投影后得到的结果估计非刚体的形变程度。实验结果表明,这种方法得到的解,误差较小,效果明显。
Because the deformation rigid motion estimation is a nonlinear inverse problem, 2D image sequence of non-rigid 3D characteristics identification technology is influenced by background diversity, motion complexity, data loss, feature point noise and so on. A result face feature recognition reconstruction is facing problems of inac- curate deformation degree estimate and large identification error. In order to solve these problems, this algorithm depicts situations of different rigid motion shape by adding physical motion restrict in frame model of 3 D face motion restoration, then accurately calculates the extent of the deformation face movement, use the method of solving non- linear optimization closed campaign for the reconstruction of the structural parameters of model, using the projection later got results estimate the degree of deformation rigid body. Experimental results show that this method has small errors and obvious recognition effect.
出处
《科学技术与工程》
北大核心
2013年第11期2911-2915,共5页
Science Technology and Engineering
关键词
特征运动簇
非线性优化
形变分析
characteristics of movement clusters nonlinear optimization deformation analysis
作者简介
第一作者简介：蒋琳琼（1975-），女，汉族，湖南永州人，硕士研究生，讲师。研究方向：智能信息处理，嵌入式系统，物联网技术。
