高级检索

基于Kanade-Lucas-Tomasi 算法的人脸特征点跟踪方法

Tracking Facial Feature Points Using Kanade-Lucas-Tomasi Approach

  • 摘要: 与传统的在人面部画上标识点的特征点跟踪方法不同,KLT (Kanade-Lucas-Tomasi)算法可以从未加标识点的正面人像视频系列中通过特征纹理信息直接获取面部某些特征点的位移.在KLT算法中加入了基于人脸统计信息的经验约束,使KLT算法更加合理有效.

     

    Abstract: Compared to normal methods that need to make marks on person’s face, this method can track displacements of facial feature points from frontal face video through texture information of the face. At the same time, in order to improve the method, we give the static angle and distance limitation of facial organs to the KLT(Kanade-Lucas-Tomasi) algorithm.

     

/

返回文章
返回