摘要
针对传统的基于单一特征的跟踪方法在复杂场景和光照变化下易导致跟踪失败的缺点,提出了一个基于多特征自适应融合的目标跟踪算法。首先选取具有互补性的目标颜色和纹理特征构造目标的多特征模型;然后根据特征子模型对目标与背景的可分性,对目标特征子模型的权值进行自适应调节;最后利用颜色和纹理特征对所提的算法进行了验证。试验表明同基于单个特征的核函数目标跟踪方法相比具有更好的鲁棒性。
Traditional tracking methods based on a single characteristic in complex scenes and in the case of the illumination change easily lead to tracking failed. Therefore this paper presented a target tracking algorithm on multi-feature adaptive fu- sion. Firstly this paper constructed target multi-feature model by using complementary color and texture characteristics of the tar- get; Secondly according to the separability of target and background of feature submodel, adjusted adaptively weight of target fea- ture submedel; Finally, using both color and texture features verified the algorithm proposed in this paper. The tests indicate that, target tracking algorithm on multi-feature adaptive fusion is more robustness.
出处
《华北科技学院学报》
2012年第3期23-27,共5页
Journal of North China Institute of Science and Technology
基金
华北科技学院校立基金资助项目(2011B031):"中央高校基本科研业务费"资助项目
关键词
多特征
颜色
纹理
权值
multi-feature
color
texture features
weight