Target tracking of the scale and rotation problem research

In order to solve the scale and rotation problem in target tracking, a target tracking algorithm based on scale invariant feature transform and mean shift is proposed. First, template area and the target area in the extreme value point of scale space are detected, then accurate position and scale of the feature point are detected through the simulation of three dimensional quadratic function, after that feature points of the target area and template area are matched, and error matching is removed according to the continuity of scale and angle between adjacent frame. The experimental results show that the algorithm has good tracking performance when targets angle and scale changed.