摘要
针对运动目标跟踪过程中出现的遮挡问题,提出了基于目标先验信息的视觉显著性遮挡目标跟踪算法。在粒子滤波框架下,利用目标先验信息生成视觉显著图,并根据粒子区域颜色特征与目标颜色特征模板之间的相似度来判断遮挡情况。当遮挡发生时,提高特征融合公式中显著性特征的融合权重,从而充分利用目标未被遮挡部分信息来完成跟踪。实验结果表明,利用目标先验信息的目标跟踪算法能显著提升跟踪遮挡目标的鲁棒性。
To address the occlusion problem during target tracking,a particle filter tracking algorithm based on visual saliency and priori information of the target was presented. In the framework of particle filter,the visual saliency map is produced using the priori information of the target and the occlusion situation is decided on the similarity between the color feature in the particle area and the target area. When occlusion happens,the weight of the saliency feature is increased in the feature fusion formula,so that the object can be located using the information of the part that not be occluded on the object. Experimental results show that the algorithm can effectively deal with the situation when target is occluded.
出处
《微型机与应用》
2016年第4期46-49,共4页
Microcomputer & Its Applications
关键词
目标跟踪
先验信息
粒子滤波
显著性特征
object tracking
priori information
particle filter
saliency feature
