基于稀疏表示的目标追踪算法

Object Tracking Algorithm Based on Sparse Representation

针对目标在运动过程中会发生光照和姿态变化,背景干扰,遮挡等因素造成目标偏移甚至丢失的问题,提出了一种基于稀疏表示的目标追踪算法。在粒子滤波框架下,采用目标在超完备字典下的稀疏表示作为观测模型,通过l1范数优化求解稀疏表示系数,并利用重构残差更新粒子权重,可以有效地减小背景杂波和噪声对追踪算法的不利影响。实验结果表明,该算法有较好的稳定性和鲁棒性。

In order to resolve the drift and even to reduce problems when objects undergo illumination and pose change, clutter, occlusion, we present an object tracking algorithm based on sparse representation. We use the sparse represen- tation under the over complete dictionary as the observation model in a particle filter framework. The optimal sparse coding coefficients are obtained via It norm minimization and the weight of the particle is updated by the re-construction en-or. Our tracker can reduce the adverse impact of noise and background clutter. Experimental results show that the algorithm is more effective and robust than the existing method.