Color-Based Object Tracking Using Fast Survival-of-the-Fittest Method

The efficient tracking of visual objects in complex environments is a challenging task for the vision community. In a real-rime particle filter tracking, the particles are a constrained resource and should be well used. From the aspect of efficiently exploring the state space, a new particle filter method named detection based particle filter to the real time tracking of non- rigid objects are proposed. This method evenly spread particles in the possible state space at each time step, and keeps a minimum particle number in each possible target region. Color clues are used in this method. Experimental results show that this method can achieve good performance and need smaller particle number compared with sampling importance resampling (SIR) particle filter.