An Improved Correlation Filter-Based Target Tracking Method

Visual tracking is still a challenging task as the objects suffer significant appearance changes, fast motion, and serious occlusion in practical applications. In this paper, we propose an improved correlation filter-based tracking method for robust object tracking. We first propose a criterion that combines the response value and the object displacement difference to determine whether the tracker drifts. Then, in the re-detector, the Edge boxes method is used to detect potential objects. When the criterion is triggered, the re-detection mechanism is executed and the tracker is commanded to stop, and then the re-detector selects the most reliable object to reinitialize the tracker. Extensive experiments demonstrate that the proposed method is superior to some of the most advanced methods in terms of accuracy, speed and robustness.