摘要
为改善粒子退化现象,提高粒子滤波算法对于遮挡目标的跟踪精度,提出加权颜色特征与SIFT特征双层融合的粒子滤波算法。用粒子的位置信息对传统粒子权重进行改进,作为内层融合,对SIFT点匹配跟踪框与粒子滤波跟踪框进行外融合,改善遮挡情况下粒子退化的情况,提高跟踪精度。在VOT2014视频数据库上的实验结果表明,提出算法平均跟踪精度为95.03%,较传统粒子滤波算法提高了6个百分点。
Particle degeneracy is a serious problem for occlusion object tracking in terms of particle filter algorithm.To solve the problem,double combination of weighted color feature and SIFT based particle filter algorithm was proposed.The location information was adopted to improve the conventional particle weight as the inner fusion.The tracking box of SIFT points matching and that of particle filter were fused afterwards to improve particle degeneracy and promote the tracking accuracy.Experimental results on VOT2014 video database show that the average tracking accuracy of the proposed algorithm is 95.03%,which is 6 percentage points higher than that of the traditional particle filter algorithm.
作者
魏旭东
秦立峰
WEI Xu-dong;QIN Li-feng(College of Mechanical and Electronic Engineering,Northwest A&F University,Yangling 712100,China;School of Electronic and Information Engineering,Xi'an Jiaotong University,Xi'an 710049,China;Key Laboratory of Agricultural Internet of Things,Ministry of Agriculture,Yangling 712100,China)
出处
《计算机工程与设计》
北大核心
2019年第2期556-561,共6页
Computer Engineering and Design
基金
陕西省农业科技创新与攻关基金项目(2015NY034)
关键词
目标跟踪
遮挡
粒子滤波
SIFT点
算法融合
粒子退化
visual tracking
occlusion
particle filter
SIFT point
algorithm combination
particle degeneracY
