结合水平集和粒子滤波的人脸轮廓跟踪

针对视频序列中的人脸轮廓跟踪问题,提出了一种将改进水平集方法引入到粒子滤波框架中的轮廓跟踪算法.此算法使用零水平集函数对人脸轮廓进行表示,通过水平集函数的演化对人脸轮廓进行逼近.为了解决人脸遮挡问题,将形状先验加入到演化过程中来约束曲线的演化.算法在粒子滤波的框架下可以同时跟踪人脸区域的仿射运动和人脸轮廓的形变.实验结果表明,所提出的'方法可以对人脸轮廓进行精确的跟踪,并对外界的光照变化,背景干扰,人脸部分遮挡有较强的鲁棒性.

Face Contour Tracking by Combing Level Set Method and Particle Filtering

Aimed at the human face contour tracking in video image sequences,a contour tracking algorithm is proposed in which an improved level set method is introduced into the particle filtering framework.The face contour is represented by zero level set function and approached by evolution of the level set function.To deal with the partial occlusion,the shape prior is added to the updating process to constrain the curve evolution.This algorithm can simultaneously track the afline motion of the face region and deformation of the face contour in the framework of particle filtering.Experimental results show that the proposed algorithm can track the face contour accurately and be robust to illumination changes,background interference and partial occlusion.