Target tracking using corner matching

The corner feature is the important feature of an image. The corner feature needs small calculated amount and can be matched easily. Moreover, the corner feature can keep the invariance after being rotated, translated or scaled. Therefore, it plays an important role in the application fields such as image registration and matching, object recognition, motion analysis and target tracking. A new method of vector matching based on the corner feature is presented in this paper. Firstly, a Harris operator is used to extract the corners of a target, and then the feature vector of the target comes out from the moment feature of corners. Finally, the match feature vector in the sequence images can be used to track the target. Using this method, more than 80% of corners in the normal condition and 70% of corners in the shelter conditions can be matched. Processing speed can achieve 20 frames/s to meet the real-time requirements. Experiment result shows that this method can effectively resist the deformation of the target, shelter and so on.