動画像に対して運動量や深度値から主被写体を決定することで,高精度な動画像クロッピング手法を実現した.また,被写体の動きを考慮したシーム算出により,位置ずれと歪みを抑制した高精度なSCを実現した.今後は,これらの手法の更なる精度向上および融合により,より汎用的なリサイズ手法に拡張する.
【参考文献】
[1] Richard Droste, Jianbo Jiao, J. Alison Noble, “Unified Image and Video Saliency Modeling”, In Proceedings of the European Conference on Computer Vision (ECCV), 2020.
[2] Kao Zhang, Yan Shang, Songnan Li, Shan Liu, Zhenzhong Chen, “SalCrop: Spatio-temporal Saliency Based Video Cropping”, IEEE International Conference on Visual Communications and Image Processing (VCIP) demo paper, 2022.
[3] Adobe, “Automatically reframe video for social media channels”, <https:// helpx.adobe.com/premiere-pro/using/auto-reframe.html>
[4] S.Wang, et al. “Multi-Operator Video Retargeting Method Based on Improved Seam Carving”, 2020 IEEE 5th Information Technology and Mechatronics Engineering Conference (ITOEC), pp.1609-1614, 2020.
[5] Shoma Deguchi, et al. “Fast Seam Carving for Video Image Based on Cost Equalisation”, Proceedings of 2022 IEEE 11th Global Conference on Consumer Electronics (GCCE), 2022.