An enhanced seam carving approach for video retargeting

Video retargeting (resizing) is an important task for displaying videos on various display devices. In this study, an enhanced seam carving approach for video retargeting is proposed, in which a seam may be a non-8-connected one. Both the search window size and the temporal weight can be adaptively adjusted according to video contents (motion information). Additionally, to preserve temporal coherence, the appearance-based method is employed. The spatial and temporal costs of a pixel are linearly combined to compute the cumulative cost with an adaptive temporal weight. Finally, dynamic programming is used to determine the optimal non-8-connected seam (with the minimum cumulative cost) for carving out. Based on the experimental results obtained in this study, the performance of the proposed approach is better than those of two comparison approaches.