基于粒子滤波框架联合仿射和外貌模型的目标跟踪

针对视频序列中目标的跟踪问题,提出了一种基于粒子滤波框架的联合仿射和外貌模型的目标跟踪算法.该算法首先提取图像帧之间的相关特征点,通过求解Sylvester方程得到仿射参数,然后将仿射参数嵌入到基于仿射群的粒子滤波框架中进行平滑估计.利用基于仿射群的一阶自回归过程模拟状态的变化,联合仿射特征点模型和外貌模型进行似然估计,得到粒子的最佳平均状态,进而对目标实施跟踪.实验结果表明,在目标经历姿势和尺度变化、遮挡以及复杂背景等情况下,提出的算法能够有效地跟踪目标,较之其他相关算法具有很强的鲁棒性.

Object tracking based on particle filtering framework joint affine model and appearance model

An algorithm based on particle filtering framework joint affine model and appearance model is introduced in this paper for object tracking in video sequences.The affine parameters for pose estimation from the corresponding feature points,which are extracted between two successive images,can be formed as a solution to Sylvester’s equation.Then,the affine parameters can be smoothly estimated within the particle filtering framework based on the affine group.The state dynamic is modeled via the first order autoregressive(AR) process on the affine group.And the optimal mean state of particles is estimated through the total likelihood function which is a combination of affine feature model and appearance model.Thus,object tracking can be actualized via particle filtering based on the affine group.Experimental results demonstrate the proposed method is more effective and robust compared with other algorithms when the tracked object undergoes pose and scale changes,occlusion and complex background.