Particle filter tracking method of color and SIFT features based adaptive fusion.

In order to solve the problem of motion blur and partial occlusion during the object tracking in the sequence images, this paper proposed a particle filter tracking method based on the improved color histogram model and the scale invariant feature transform (SIFT) model. According to the track environment, it adaptively allocated the weights of the two types of target information by using fuzzy logic method, thus improved the reliability of target observation description by feature fusion. Experimental results show that the proposed method is superior to the traditional methods which is based on only one feature or fixed weight. [ABSTRACT FROM AUTHOR]