F. Mohanna and F. Mokhtarian (UK)
corner detection, curvature scale space, Canny edge detector, corner matching, correspondence
This paper presents a multi-scale corner tracking algorithm based on a multi-scale corner detector and two-frame matching. To extract corners from each frame of video sequence, the enhanced CSS1 corner detector using different scales of smoothing is applied. In two-frame correspondence, for each tracked corner in current frame, a single match is considered in next frame using standard cross-correlation metric. Problem of recovering points lost during tracking is solved considering three-frame based monitoring. We monitor tracked corners from the third frame of input sequence due to occlusion or sudden change in drift of the tracker. As a result, proposed three-frame monitoring helps the new tracked corners to be added to the list of tracked corners in test sequence. Since the proposed corner tracker has enough robust corners based on the multi-scale corner detector, it is practical and efficient. Experiments have been carried out for a wide range of real video databases depicting translation, scaling, rotation and affine transformation with different lighting and different camera motions. All experiments confirm that the performance of the proposed tracker is reliable due to monitoring matched corners among frames.
Important Links:
Go Back
