改进的梯度与肤色融合均值移动粒子滤波人脸跟踪

Face tracking based on improved fusion of gradient and color mean shift particle filter

针对单一肤色特征的跟踪算法鲁棒性不高的问题,提出一种多特征融合的均值移动粒子滤波(MSPF)跟踪算法.该算法用肤色特征与梯度特征表示候选目标,通过粒子滤波与均值移动算法进行特征融合.实验结果表明,该算法能够较好地提高跟踪效率,并对光照、人脸遮挡和人脸旋转等有一定的适应性.

Object tracking by using single color feature results in a problem of low robustness.Mean shift particle filter（MSPF）algorithm based on multi-features fusion is proposed in this paper.The proposed method uses the color and gradient features to represent the candidate target,utilizes feature fusion in mean-shift and particle filter algorithms,and efficiently reduces the problem of particle degradation and tracking instability.Experimental results indicate the algorithm can improve the tracking accuracy,and have good adaptability to the illumination,face occlusion and rotation.