A Study On Object Tracking Based On Scale Invariant Feature Transform Features

With the development of software and hardware, the technology of object tracking has been applied in the fields of military affairs, traffic control, intelligent monitoring, GPS navigation etc. The researches of related algorithms have deepened in recent years.As particle filter has superiority in the state estimation of non-linear, non-Gaussian system, it has been wide applied in object tracking. The observation model's establishment is one of key aspects. Color based observation model is very popular, but this kind model is not robust in complex background.With the development of scale space theory, the feature matching which based on the scale space theory is robust and the combination with particle filter has become a new research hot spot in object tracking. This paper introduces the principles of scale invariant feature transform and particle filter. PF-SIFT tracking algorithm is also proposed in this paper. This algorithm integrates the scale invariant feature transform in the framework of particle filter. It has effective template choice and updating strategy to sharpen the ability of the auto-adapted of algorithm.