Super Resolution для сжатого видео
0 likes1,786 views
The document discusses super resolution techniques for compressed video. It introduces example-based super resolution, where high resolution image patches are learned from example image pairs and used to reconstruct high resolution frames. Keyframe-based super resolution is also covered, which uses high resolution keyframes as references to enhance intermediate low resolution frames. Quantitative and visual results demonstrate improved quality over baseline techniques for compressed video super resolution.
More Related Content
More from MSU GML VideoGroup (20)
Recently uploaded (20)
Super Resolution для сжатого видео
- 1. Super Resolution для сжатого видео Моисейцев Алексей Video Group CS MSU Graphics & Media Lab
- 2. Only for Maxus Содержание Введение Example-based SR SRME HMRF SR 2 CS MSU Graphics & Media Lab (Video Group)
- 3. Only for Maxus Введение 3 CS MSU Graphics & Media Lab (Video Group)
- 4. Only for Maxus Введение Lossless 2Mbps 1Mbps 256Kbps Bi-cubic source Super-res Super-Resolving Compressed Video with Large Artifacts, Wen-Yi Zhao, ICPR 4 CS MSU Graphics & Media Lab (Video Group) 2004
- 5. Only for Maxus Введение Алгоритмы SR: Iterative Backprojection (IBP) Projection Onto Convex Sets (POCS) Probabilistic Methods Maximum a posteriori (MAP) Model-based approach (MBSR) Example-based 6 CS MSU Graphics & Media Lab (Video Group)
- 6. Only for Maxus Введение IBP POCS C2 C1 C f=Pg P1g P2g C3 P3g g 7 CS MSU Graphics & Media Lab (Video Group)
- 7. Only for Maxus Введение 8 CS MSU Graphics & Media Lab (Video Group) Super resolution: an overview, C Papathanassiou and M Petrou, Geoscience and Remote Sensing Symposium, 2005
- 8. Only for Maxus Введение Типичная скорость работы: 0.15-0.4 fps 9 CS MSU Graphics & Media Lab (Video Group)
- 9. Only for Maxus Содержание Введение Example-based SR SRME HMRF SR 10 CS MSU Graphics & Media Lab (Video Group)
- 10. Only for Maxus Example-based SR В SR только по LR-кадрам существует теоретический предел качества Иногда есть доступ и к отдельным кадрам в высоком разрешении 11 CS MSU Graphics & Media Lab (Video Group) Resolution enhancement of lowKlamer Schutte,using a high-resolution frame, Tuan Phama, Lucas van Vlieta, quality videos SPIE 2006
- 11. Only for Maxus Example-based SR Registration Для компенсации движения используется алгоритм Lucas-Kanade: 12 CS MSU Graphics & Media Lab (Video Group) Resolution enhancement of lowKlamer Schutte,using a high-resolution frame, Tuan Phama, Lucas van Vlieta, quality videos SPIE 2006
- 12. Only for Maxus Example-based SR Registration В 99% случаев ошибка не превышает 0.15 LR-пикселя и 0.45 HR-пикселя 13 CS MSU Graphics & Media Lab (Video Group) Resolution enhancement of lowKlamer Schutte,using a high-resolution frame, Tuan Phama, Lucas van Vlieta, quality videos SPIE 2006
- 13. Only for Maxus Example-based SR Построение пар HR-LR Используются LR-блоки 8x8 Рассматриваются повороты блоков Ключ — 10 AC-коэффициентов LR- блока и 68 граничных пикселей HR-блока 14 CS MSU Graphics & Media Lab (Video Group) Resolution enhancement of lowKlamer Schutte,using a high-resolution frame, Tuan Phama, Lucas van Vlieta, quality videos SPIE 2006
- 14. Only for Maxus Example-based SR Восстановление 15 CS MSU Graphics & Media Lab (Video Group) Resolution enhancement of lowKlamer Schutte,using a high-resolution frame, Tuan Phama, Lucas van Vlieta, quality videos SPIE 2006
- 15. Only for Maxus Example-based SR Восстановление 16 CS MSU Graphics & Media Lab (Video Group) Resolution enhancement of lowKlamer Schutte,using a high-resolution frame, Tuan Phama, Lucas van Vlieta, quality videos SPIE 2006
- 16. Only for Maxus Example-based SR Результат source, frame 0 compressed frame 20 SR, frame 20 MJPEG, Q=50 Обучение на первом кадре Восстановление двадцатого кадра 17 CS MSU Graphics & Media Lab (Video Group) Resolution enhancement of lowKlamer Schutte,using a high-resolution frame, Tuan Phama, Lucas van Vlieta, quality videos SPIE 2006
- 17. Only for Maxus Example-based SR Результат compressed frame 20 SR, frame 20 Spatial example-based SR, frame 20 18 CS MSU Graphics & Media Lab (Video Group) Resolution enhancement of lowKlamer Schutte,using a high-resolution frame, Tuan Phama, Lucas van Vlieta, quality videos SPIE 2006
- 18. Only for Maxus Key-frame based SR Идея: использовать HR ключевые кадры 19 CS MSU Graphics & Media Lab (Video Group) Super Resolution ofMukherjee, ISCASFrames, Fernanda Brandi, Ricardo de Queiroz, Debargha Video Using Key 2008
- 19. Only for Maxus Key-frame based SR Восстановление 20 CS MSU Graphics & Media Lab (Video Group) Super Resolution ofMukherjee, ISCASFrames, Fernanda Brandi, Ricardo de Queiroz, Debargha Video Using Key 2008
- 20. Only for Maxus Key-frame based SR Результаты 21 CS MSU Graphics & Media Lab (Video Group) Super Resolution ofMukherjee, ISCASFrames, Fernanda Brandi, Ricardo de Queiroz, Debargha Video Using Key 2008
- 21. Only for Maxus Key-frame based SR Результаты 22 CS MSU Graphics & Media Lab (Video Group) Super Resolution ofMukherjee, ISCASFrames, Fernanda Brandi, Ricardo de Queiroz, Debargha Video Using Key 2008
- 22. Only for Maxus Key-frame based SR Результаты 23 CS MSU Graphics & Media Lab (Video Group) Super Resolution ofMukherjee, ISCASFrames, Fernanda Brandi, Ricardo de Queiroz, Debargha Video Using Key 2008
- 23. Only for Maxus Example-based SR Слабая обоснованность метода Требуется наличие специфичного видеопотока Зависимость от обучающей выборки 24 CS MSU Graphics & Media Lab (Video Group)
- 24. Only for Maxus Содержание Введение Example-based SR SRME HMRF SR 25 CS MSU Graphics & Media Lab (Video Group)
- 25. Only for Maxus SRME Simultaneous motion estimation and resolution enhancement of compressed low resolution video. 26 CS MSU Graphics & Media Lab (Video Group) Javier Mateos, A. K. Katsaggelos, Rafael Molina, ICIP, 2000
- 26. Only for Maxus SRME Simultaneous motion estimation and resolution enhancement of compressed low resolution video. 27 CS MSU Graphics & Media Lab (Video Group) Javier Mateos, A. K. Katsaggelos, Rafael Molina, ICIP, 2000
- 27. Only for Maxus SRME — апостериорная вероятность — интересующее решение Simultaneous motion estimation and resolution enhancement of compressed low resolution video. 28 CS MSU Graphics & Media Lab (Video Group) Javier Mateos, A. K. Katsaggelos, Rafael Molina, ICIP, 2000
- 28. Only for Maxus SRME Задание вероятностей (1) Simultaneous motion estimation and resolution enhancement of compressed low resolution video. 29 CS MSU Graphics & Media Lab (Video Group) Javier Mateos, A. K. Katsaggelos, Rafael Molina, ICIP, 2000
- 29. Only for Maxus SRME Задание вероятностей (2) Simultaneous motion estimation and resolution enhancement of compressed low resolution video. 30 CS MSU Graphics & Media Lab (Video Group) Javier Mateos, A. K. Katsaggelos, Rafael Molina, ICIP, 2000
- 30. Only for Maxus SRME Задание вероятностей (3) Simultaneous motion estimation and resolution enhancement of compressed low resolution video. 31 CS MSU Graphics & Media Lab (Video Group) Javier Mateos, A. K. Katsaggelos, Rafael Molina, ICIP, 2000
- 31. Only for Maxus SRME Решение (1) Simultaneous motion estimation and resolution enhancement of compressed low resolution video. 32 CS MSU Graphics & Media Lab (Video Group) Javier Mateos, A. K. Katsaggelos, Rafael Molina, ICIP, 2000
- 32. Only for Maxus SRME Решение (2) Две итерации: Компенсация движения Построение улучшенного кадра Simultaneous motion estimation and resolution enhancement of compressed low resolution video. 33 CS MSU Graphics & Media Lab (Video Group) Javier Mateos, A. K. Katsaggelos, Rafael Molina, ICIP, 2000
- 33. Only for Maxus SRME source Y-PSNRSimultaneous motion estimation and resolution enhancement of = 20.44dB compressed low resolution video. 34 CS MSU Graphics & Media Lab (Video Group) Javier Mateos, A. K. Katsaggelos, Rafael Molina, ICIP, 2000
- 34. Only for Maxus SRME bilinear Y-PSNRSimultaneous motion estimation and resolution enhancement of = 21.98dB compressed low resolution video. 35 CS MSU Graphics & Media Lab (Video Group) Javier Mateos, A. K. Katsaggelos, Rafael Molina, ICIP, 2000
- 35. Only for Maxus SRME proposed SR Y-PSNRSimultaneous motion estimation and resolution enhancement of = 25.64dB compressed low resolution video. 36 CS MSU Graphics & Media Lab (Video Group) Javier Mateos, A. K. Katsaggelos, Rafael Molina, ICIP, 2000
- 36. Only for Maxus Содержание Введение Example-based SR SRME HMRF SR 37 CS MSU Graphics & Media Lab (Video Group)
- 37. Only for Maxus HMRF SR Standard decompression Smoothed image Sharpened image Restoration of Compressed Video using Temporal Information, 38 CS MSU Graphics & Media Lab (Video Group) Mark A. Robertson and Robert L. Stevenson, SPIE 2001
- 38. Only for Maxus HMRF SR Restoration of Compressed Video using Temporal Information, 39 CS MSU Graphics & Media Lab (Video Group) Mark A. Robertson and Robert L. Stevenson, SPIE 2001
- 39. Only for Maxus HMRF SR Шум квантования (1) Restoration of Compressed Video using Temporal Information, 40 CS MSU Graphics & Media Lab (Video Group) Mark A. Robertson and Robert L. Stevenson, SPIE 2001
- 40. Only for Maxus HMRF SR Шум квантования (2) Restoration of Compressed Video using Temporal Information, 41 CS MSU Graphics & Media Lab (Video Group) Mark A. Robertson and Robert L. Stevenson, SPIE 2001
- 41. Only for Maxus HMRF SR Шум компенсации (1) Restoration of Compressed Video using Temporal Information, 42 CS MSU Graphics & Media Lab (Video Group) Mark A. Robertson and Robert L. Stevenson, SPIE 2001
- 42. Only for Maxus HMRF SR Шум компенсации (2) — общее выражение — независимость шума Restoration of Compressed Video using Temporal Information, 43 CS MSU Graphics & Media Lab (Video Group) Mark A. Robertson and Robert L. Stevenson, SPIE 2001
- 43. Only for Maxus HMRF SR Условие гладкости Restoration of Compressed Video using Temporal Information, 44 CS MSU Graphics & Media Lab (Video Group) Mark A. Robertson and Robert L. Stevenson, SPIE 2001
- 44. Only for Maxus HMRF SR — гладкость — квантование — компенсация Restoration of Compressed Video using Temporal Information, 45 CS MSU Graphics & Media Lab (Video Group) Mark A. Robertson and Robert L. Stevenson, SPIE 2001
- 45. Only for Maxus HMRF SR Решение Градиент: Restoration of Compressed Video using Temporal Information, 46 CS MSU Graphics & Media Lab (Video Group) Mark A. Robertson and Robert L. Stevenson, SPIE 2001
- 46. Only for Maxus HMRF SR Решение Упрощение: Градиентный спуск: Restoration of Compressed Video using Temporal Information, 47 CS MSU Graphics & Media Lab (Video Group) Mark A. Robertson and Robert L. Stevenson, SPIE 2001
- 47. Only for Maxus HMRF SR Результаты Restoration of Compressed Video using Temporal Information, 48 CS MSU Graphics & Media Lab (Video Group) Mark A. Robertson and Robert L. Stevenson, SPIE 2001
- 48. Only for Maxus HMRF SR Результаты Restoration of Compressed Video using Temporal Information, 49 CS MSU Graphics & Media Lab (Video Group) Mark A. Robertson and Robert L. Stevenson, SPIE 2001
- 49. Only for Maxus HMRF SR Результаты Restoration of Compressed Video using Temporal Information, 50 CS MSU Graphics & Media Lab (Video Group) Mark A. Robertson and Robert L. Stevenson, SPIE 2001
- 50. Only for Maxus Заключение Рассмотрены Example-based и MAP методы Многие методы используют информацию из потока или требуют дополнительные данные для работы Прямые реализации алгоритмов сложны, но тем не менее многие из них возможно распараллелить (например, с использованием CUDA) 51 CS MSU Graphics & Media Lab (Video Group)
- 51. Only for Maxus Литература Simultaneous motion estimation and resolution enhancement of compressed low resolution video. Javier Mateos, A. K. Katsaggelos, Rafael Molina, ICIP, 2000 Restoration of Compressed Video using Temporal Information, Mark A. Robertson and Robert L. Stevenson, SPIE 2001 DCT Quantization Noise in Compressed Images, Mark A. Robertson and Robert L. Stevenson, 2004 Super-Resolving Compressed Video with Large Artifacts, Wen-Yi Zhao, ICPR 2004 Z. Lin and H-Y. Shum. Fundamental limits of reconstruction-based superresolution algorithms under local translation. PAMI, 2004 Resolution enhancement of low quality videos using a high-resolution frame, Tuan Phama, Lucas van Vlieta, Klamer Schutte, SPIE 2006 T.Q. Pham, M. Bezuijen, L.J. van Vliet, K. Schutte, and C.L. Luengo Hendriks. Performance of optimal registration estimators, SPIE 2005 Super resolution: an overview, C Papathanassiou and M Petrou, Geoscience and Remote Sensing Symposium, 2005 Super Resolution of Video Using Key Frames, Fernanda Brandi, Ricardo de Queiroz, Debargha Mukherjee, ISCAS 2008 52 CS MSU Graphics & Media Lab (Video Group)
- 52. Only for Maxus Вопросы ? 53 CS MSU Graphics & Media Lab (Video Group)