自助重要性采样用于实时多目标视觉跟踪  

Real-time Visual Tracking of Multiple Targets Using Bootstrap Importance Sampling

作　　者：沈乐君[1,2] 游志胜[1,3] 李晓峰[1,3] 

机构地区：[1]四川大学视觉合成图形图像技术国防重点学科实验室 ,成都 610064 [2]3. 四川大学计算机学院 ,成都 610064 [2]成都体育学院信息技术中心 ,成都 610041 [3]四川大学计算机学院,成都610064

出　　处：《自动化学报》2012年第10期1663-1670,共8页Acta Automatica Sinica

基　　金：国家高技术研究发展计划(863计划)(2012AA011804);国家自然科学基金(61001195);四川省科技支撑计划(2010JY0078)资助~~

摘　　要：多目标视觉跟踪的主要困难来自于多个目标交互(部分或完全遮挡)导致的歧义性.马尔可夫随机场(Markov random field,MRF)可以消除这种歧义性且无需显式的数据关联.但是,通用概率推理算法的计算代价很高.针对上述问题,本文做出了3点贡献:1)设计了新的具有"分散-集中-分散"结构的递归贝叶斯跟踪框架—自助重要性采样粒子滤波器,它使用融入当前时刻观测的重要性密度函数解决维数灾难问题,将计算复杂度从指数增长变为线性增长;2)提出了新的蒙特卡洛策略—自助重要性采样,利用MRF的因子分解性质进行重要性采样,并使用自助法产生低成本高质量的样本、降低似然度计算次数和维持多模式分布;3)采用了新的边缘化技术—使用辅助变量采样进行边缘化,使用自助直方图对边缘后验分布进行密度估计.实验结果表明,本文提出的算法能够对大量目标进行实时跟踪,能够处理目标间复杂的交互,能够在目标消失后维持多模式分布.Ambiguity is the major difficulty in multi-object tracking problem due to the interactions of multiple targets （partial or complete occlusion）. This ambiguity can be resolved by Markov random field （MRF） without explicit data association. However, the computational cost of generM probabilistic inference algorithms of MRF is expensive. This paper presents a novel approach to this problem. Firstly, a new recursive Bayesian estimation framework, bootstrap importance sampling particle filter （BIS-PF）, is devised, which has a ＂distributed-central-distributed＂ structure. The core of this framework is a suboptimal importance density which uses the observation at present time. So, it does not suffer from the curse of dimensionality. Secondly, a new Monte Carlo strategy is proposed, which uses bootstrap sampling to generate low- cost and high-quality samples, maintains multi-modality and decreases the number of likelihood computations. Thirdly, a new marginalization technology is presented, which uses an auxiliary variable sampler to obtain marginal samples and bootstrap based histogram for density estimation. The experiments show that the proposed method can track multiple targets in real-time, handle the complex interaction and maintain multi-modalities even the objects disappear.

关 键 词：多目标跟踪 视觉跟踪 粒子滤波 马尔可夫随机场 自助法 重要性采样 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]