基于Kinect与网格几何变形的人脸表情动画

摘要

The efficiency and realistic of facial expression reconstruction is a key point to facial expression animation. A novel method of rapidly facial expression reconstruction based on face tracking with Kinect and geometry deformation has been proposed. The face of actor has been identified using the Kinect device of Microsoft and the feature points have been recorded. These feature points are used to construct a mesh model covering the actor's face. Some control points have been chosen for mesh deformation from the mesh model. Because of the real-time face tracking of actor the rapid reconstruct of target model using three different deformation algorithm is realized. The experimental results show that this method is easily implemented and needn't any calibration, automatically immigrates the facial expression to target model. The expression of result model is true and natural.%面部表情重建的实时性与重建效果的真实性是人脸表情动画的关键问题,提出一种基于Kinect人脸追踪和几何变形技术的面部表情快速重建新方法.使用微软的Kinect设备识别出表演者面部并记录其特征点数据,并利用捕捉到的特征点建立覆盖人脸的网格模型,从中选取变形使用的控制点数据,由于Kinect可以实时地自动追踪表演者面部,由此实现了利用三种不同变形算法对目标模型实时快速重建.实验结果表明,该方法简单易实施,不用在表演者面部做任何标定,可以自动地将人脸表情动作迁移到目标模型上,实现人脸表情快速重建,并且保证目标模型表情真实自然.