摘要: |
采用自适应人脸方向模板和YCbCr自适应肤色模型,提出了一种薪的基予趣值德移算法的自适应人脸跟踪方法。与传统的均值漂移跟踪方法相比,当人脸倾斜时或光线变化时,该方法能更精确地描绘出人脸位置。实验结果 表明,在基本上不增加计算量的情况下,该方法能对人脸的倾斜和光线的变化进行很好地自适应跟踪。 |
关键词: 均值漂移算法 自适应人脸倾角模板 人脸跟踪 |
DOI: |
分类号: |
基金项目:中国石油大学博士基金 |
|
Adaptive human face tracking based on mean shift a^rithm |
LIU Bao-di, WANG Yan-jiang, MA Hai, WANG Wu-K
|
(College of Informatmn and Control Engineering in China University Petroleumt Dongying 257061,Shandong Province Chitui)
|
Abstract: |
By using an adaptive facial orientation template and an adaptive YCbCr skin color model, a novel adaptive human face tracking method was proposed on the basis of mean shift algorithm. Compared with the traditional mean shift backing method, the method can describe the face position more accurately when facial orientation changes or lighting changes. TTie experimental results show that the method can adapt to the facial orientation and limiting variations without increasing calculated amount. |
Key words: mean shift algorithm adaptive facial orientatioti template human face tracking |