Hybrid Algorithm for Color Video Object Detection using Particle Filters

Colour can provide effective graphic features for tracking non rigid objects in real-time. However the colour of an object can vary over time dependent on the illumination, the visual angle to handle these appearance change a colour based target model must be adapted during temporally stable image observation. The proposed method of this dissertation gives new observation likelihood model with dynamic parameter setting. Experiments show our proposed method is more accurate and more efficient than the traditional colour histogram based particle filter. Integration of colour distribution into particle filters and shows how these distributions can be adopted overtime. A particle filter tracks several hypotheses simultaneously and weight them according to their similarity to the target model. As similarity measures between two colour distributions the popular Bhattacharyya coefficient is applied. In order to update the target model to slowly varying image conditions, Frames where the object is occluded or too noisy must be discarded. Keywords— Weighted Histogram, Particle Filter, Trajectory, Colour, PDF, Prediction