Improvement of Particle Filtering for Intersection of Targets with Similar Patterns
Naoki Enda, Yuji Iwahori, Robert J. Woodham · 2009
The particle filtering, which is a probability technique, can be used for a robust tracking to noise and occlusion. It often uses characteristics based on the appearance information of a target object, such as color and edge, for calculating likelihood. As far as the appearance information is used, the particle filtering fails to track the target object when it intersects other objects which have similar patterns. This paper proposes a new method for addressing the problem. The method uses color and other characteristics such as velocity information of the tracking object and distance information between tracking objects except the appearance. 1