Color-based Real-time Recognition and Tracking

Robust real-time tracking of non-rigid objects is a challenging task and is required by many vision applications such as augmented reality, smart rooms and surveillance. Particle filtering has proven very successful for non-linear and non-Gaussian estimation problems. We use color-based image features instead of the edge-based image features which have typically been used. The integration of color distributions into particle filtering has many advantages for tracking non-rigid objects as color histograms in particular are robust to partial occlusion, are rotation and scale invariant and are calculated efficiently. Based on different known histograms, objects are distinguished from each other and tracked in real-time with the proposed framework. An application is shown where the recognized objects are replaced by artificial objects.