A fast and stable seam selection algorithm for video stitching based on seam temporal propagation constraint
Peng Ling Yang, Wei Xu, Li Jing, Maojun Zhang · 2016
A fast and stable seam selection method applicable to time-synchronized video streams is proposed in this paper. The method exhibits low computational time and good performance using frame to frame correction. The main contribution of this paper is the proposed seam temporal propagation constraint that exploits spatial contextual information to avoid artifacts caused by large seam shifts between successive video frames. Furthermore, an enhanced dynamic programming algorithm is introduced to obtain optimal seams with a comparably low execution time. A video stitching platform is employed to provide a detailed experimental comparison of the proposed method with existing methodologies. Our results clearly indicate that employing contextual information in video stitching can achieve a significant reduction in artifacts.