基于特征缺失修复的人脸三维运动形变估计算法

Based on the Characteristics of the Lack of 3D Movement Repair Face Deformation Estimation

人脸由于其非刚性明显,受到运动背景变化、形变程度复杂、特征丢失等因素的影响,使得在进行三维动态人脸形变估计时,面临着估计不准,误差较大的问题。为了解决这些问题,提出一种基于特征缺失修复的人脸三维运动形变估计算法,将非刚体人脸运动的形变估计过程中,运用特征补偿方法,补偿由于噪声、形变程度复杂带来的特征丢失缺陷,运用较多的人脸形变特征进行形变分析,准确计算人脸的形变程度。实验结果表明,这种方法得到的解误差较小,效果明显。

Face because of its non-rigid obvious,background changes,deformation by sports complex, features lost de- gree the influence of factors,make during 3 d dynamic face deformation estimate,facing to estimate,error of. In order to solve these problems,the article proposes a kind of lack of repair based on feature 3 d movement face deformation esti- mation algorithm,will not face the deformation rigid body motion estimation,the use of characteristic compensation method and compensation because noise ,deformation degree complex characteristics of defects bring lost, using more face deformation characteristics of deformation analysis, the accurate calculation of deformation degree face. The experi- mental results show that this method has been of the solution of the small errors, the effect is obvious.