中文摘要：

近年来，粒子滤波算法作为一种处理非线性、非高斯问题的工具，在目标跟踪领域取得了广泛的应用．在粒子滤波中，构建一个合适的似然模型对跟踪的准确性起着关键性的作用．本文根据人体头部旋转的特点，提出了一种基于多信息融合的人脸跟踪算法．首先利用相邻两帧之间目标的位置相差很小这一信息，排除掉一些粒子是目标的可能性，然后融合颜色和局部灰度均值对比度这两个特征，在粒子滤波框架内对人脸进行跟踪．实验结果表明，该算法能够对人脸进行实时鲁棒的跟踪，并能够很好地解决在人体头部发生旋转情况下的跟踪问题．

英文摘要：

In recent years, particle filter algorithm, as a powerful tool to handle nonqinear, non Gaussian problems, has been widely used in the field of target tracking. In the particle filter algorithm, build-ing an appropriate likelihood model is crucial to the accuracy of target tracking. A novel multifeature fusion tracking algorithm is proposed based on the rotation characteristic of human head. Firstly, the position information of two adjacent frames is considered. Secondly, two features namely color model and local contrast mean difference of the candidate targets are calculated and adopted to the particle filter. Experimental results show that our methods can track human face successfully, especially when the head is rotating around.