Optic Flow Target Tracking Method Based on Corner Detection
Hua-jun Song, Mei-li Shen · 2009
In target tracking system, speed and accuracy are two important evaluations indicators. This paper uses Harris algorithm to extract little part of target corner features which can improve this method's speed. Then, it uses optic flow method which has high matching accuracy to match those extracted corner features in the following video frames. However, when the tracked target is rotated or transformation, the distributing of matched feature point is changed. In order to overcome this shortcoming, the proposed method uses bary algorithm to compute the bary of those matched features as displacement of target. It is proved by experiments that this method has not only high speed and tracking accuracy but also higher robustness to target rotation, transformation, shelter etc.