Visual tracking with re-detection based on feature combination

An improved visual tracking method is proposed, which combines the image gradient feature and color feature, and uses correlation filter framework to undergo object tracking. This algorithm uses the Bayesian theory to model the color information, integrates gradient feature's correlation filter output and object confidence integral map obtained from dense object posterior probability for object tracking. Furthermore, this algorithm also evaluates the tracking results, the object redetection process is enabled as the tracking quality is unreliable, which uses the object confidence integral to determine the candidate objects. When the tracking quality is unreliable and no reliable object is re-detected in some frame, model learning shall be bypassed. Experiments show that its tracking precision is obviously improved compared with that of several state-of-the-arts correlation filters, which can avoid the problem of object loss and model drift due to occlusion.