Small target tracking in background using Saliency-based particle filter

摘要

This paper proposes a new particle filter algorithm which can better achieve target tracking. Due to the objects which have similar colors or motions with the target in the background area, it is easily to be affected when tracking with the target and then caused loss of the target. Therefore, the saliency based particle filter is proposed. This method mainly calculates the saliency degree of color hues and motion vectors, and uses them as the weight of the updated particles in the process of particle filter, so that the influence of the objects in the background areas with similar color or motion are suppressed, and the movement of the target can be highlighted, and finally the system can carry out precise tracking.