摘要
提出了一种新的自适应特征子空间跟踪算法,该算法通过计算跟踪目标的似然来自适应调整模型更新的权重,以减小更新过程中样本误差积累导致的模型漂移。同时,跟踪算法利用多视角贝叶斯理论框架进行多视角的信息融合,并对跟踪模型进行分块处理和更新,以提高跟踪精确度。仿真结果表明,本算法比对比算法的跟踪误差更小,并能够更好地解决目标遮挡和形变等问题,从而得到精确、高效的跟踪结果。
A new adaptive subspace tracking algorithm is proposed in this paper. The algorithm updates the appearance model in subspace by using the likelihood of the sample in order to eliminate the model drift. It processes and fuses the data in distributed way on different views under the Bayesian tracking framework, and employs multi-part appearance model for matching and updating to achieve more accurate tracking result. Experiments show that the proposed algorithm features a smaller tracking error than the comparison algorithms especially under occlusion and appearance variation, and it can track the object effectively and accurately.
出处
《信息与电子工程》
2012年第3期319-324,共6页
information and electronic engineering
基金
国家重大专项基金资助项目(No.2011ZX03003-001-02
No.2012ZX03001007-003)
华为合作项目(No.YBWL2010190)
关键词
多视角目标跟踪
自适应子空间更新
粒子滤波
分块观测模型
multi-view object tracking
adaptive subspace update
particle filter
multi-part observation model
