摘要
利用粒子空间分布和粒子观测概率信息定义了一种目标特征的确定性度量方法,将该度量应用到传统的多特征融合跟踪算法中,实现了目标特征加权值的自适应调整,使得不同场景及外部干扰条件下各种目标特征信息对跟踪结果的贡献达到最优。试验表明,基于确定性度量的跟踪算法较传统的单一特征跟踪及固定加权值的多特征融合跟踪算法有着更好的鲁棒性。
The certainty measurement of object feature is defined by using spatial distribution and observation proba- bility of particle. The measurement is introduced into the object tracking algorithm based on multi-feature fusion, and the weighted value of object feature is adjusted adaptively. So the object feature can adaptively adjust its contribution to tracking in different scenes and external disturbance. The results of the experiment show that the algorithm based on certainty measurement is more robust than traditional algorithm using single feature or multi-feature fusion with fixed weighted value.
出处
《激光与红外》
CAS
CSCD
北大核心
2015年第5期576-579,共4页
Laser & Infrared
关键词
确定性度量
粒子滤波
目标跟踪
certainty measurement
particle filter
target tracking
