基于最大后验概率的人脸平移与旋转跟踪方法研究

Displacement and rotation tracking for human face based on maximum posterior probability

针对目前大多数视觉跟踪算法只能够跟踪平移运动人脸目标,而不能跟踪旋转运动人脸目标这一问题,提出一种基于最大后验概率的人脸目标平移与旋转跟踪方法。在基于最大后验概率的视觉跟踪算法基础上,采用了自适应角度模板进行人脸目标旋转跟踪,与传统的视觉跟踪方法相比,当人脸发生倾斜时,该方法能更精确地得到人脸的位置和方向。实验结果表明,该方法能够有效地跟踪具有平移与旋转运动的人脸目标。

Since the most visual tracking can only track face target with displacement but not rotation,this paper presents a displacement and rotation tracking method based on maximum posterior probability.On the basis of maximum posterior probability visual tracking algorithm,this paper uses adaptive angle template to perform rotation tracking of face target.Compared with traditional visual tracking method,this method can accurately obtain the position and direction of human face when the face is slanted.The experimental results show that this method can efficiently track the face target with displacement and rotation.