Robust Mean Shift Tracking with Background Information

The background-weighted histogram (BWH) has been proposed in mean shift tracking algorithm to reduce the interference of background in target localization. However, the BWH also reduces the weight for part of complex object. Mean shift with BWH model is unable to track object with scale change. In this paper, we integrate an object/background likelihood model into the mean shift tracking algorithm. Experiments on both synthetic and real world video sequences demonstrate that the proposed method could effectively estimate the scale and orientation changes of the target. The proposed method can still robustly track the object when the target is not well initialized.

[1]  B. Kröse,et al.  An EM-like algorithm for color-histogram-based object tracking , 2004, CVPR 2004.

[2]  Dorin Comaniciu,et al.  Kernel-Based Object Tracking , 2003, IEEE Trans. Pattern Anal. Mach. Intell..

[3]  Robert B. Ash,et al.  Information Theory , 2020, The SAGE International Encyclopedia of Mass Media and Society.

[4]  Larry D. Hostetler,et al.  The estimation of the gradient of a density function, with applications in pattern recognition , 1975, IEEE Trans. Inf. Theory.

[5]  Robert T. Collins,et al.  Mean-shift blob tracking through scale space , 2003, 2003 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, 2003. Proceedings..

[6]  Geoffrey J. McLachlan,et al.  Finite Mixture Models , 2019, Annual Review of Statistics and Its Application.

[7]  D. Zhang,et al.  Robust mean-shift tracking with corrected background-weighted histogram , 2012 .

[8]  Gary Bradski,et al.  Computer Vision Face Tracking For Use in a Perceptual User Interface , 1998 .

[9]  Yizong Cheng,et al.  Mean Shift, Mode Seeking, and Clustering , 1995, IEEE Trans. Pattern Anal. Mach. Intell..

[10]  D. Zhang,et al.  Scale and orientation adaptive mean shift tracking , 2012 .