Adaptive multiple cues integration for robust outdoor vehicle visual tracking

Aiming at the robust visual tracking for outdoor vehicle, we propose an adaptive multiple cues integration tracking approach in the particle filter framework. The reliability of observation likelihood probability of each cue is estimated according to the uncertainty metric factor of each cue and the spatial distribution of particles with that cue. Then, we compute the integration weight of each cue adaptively according to the reliability of each cue and void ad-hoc tuning the integration weight. Finally, we would obtain the combination likelihood probability of all the observations in an additive integration way. We conduct tracking experiment on three sets of representative outdoor vehicle tracking video sequences to test and compare proposed adaptive multiple cues integration scheme with state-of-the-art approaches. Experimental results demonstrate that our approach is robust to vehicle-colored distracters and partial even full occlusions and outperform the classic approach in the tracking accuracy.