Multi-Person Tracking Based on Video Information Fusion
Jie Cao, Jing Run Zheng · Advanced materials research · 2011
In order to solve the problem that human objects in complex environments would be affected by illumination, deformation, moving, etc, a multi-feature information fusion based video tracking approach was proposed. In a particle filter framework, both the face color information and the head contour cue were fused effectively.This method overcomed the instability brought by using a single measurement source, and improved the tracking accuracy and reliability. Experiments on real video sequences demonstrate that in the same case our approach has more reliable performance than trackers utilizing only color or contour information with particle filter.