Abstract | ||
---|---|---|
•In the coarse-to-fine process, we conceive a bidirectional temporal feature transfer that passes the internal features of a blurry video frame on to the encoding process of its adjacent frames.•We design a novel merge subnetwork for merging extracted internal features with the direct features from adjacent frames.•We propose an innovative group encoder and hybrid decoder solution to fuse adjacent frames’ information synchronously, and obtained comparable or even superior results to other state-of-the-art methods for multi-frame video deblurring. |
Year | DOI | Venue |
---|---|---|
2020 | 10.1016/j.neucom.2019.09.031 | Neurocomputing |
Keywords | DocType | Volume |
Blind motion deblurring,Triple adjacent frames,Temporal feature transfer,Group encoder,Hybrid decoder | Journal | 376 |
ISSN | Citations | PageRank |
0925-2312 | 0 | 0.34 |
References | Authors | |
0 | 6 |
Name | Order | Citations | PageRank |
---|---|---|---|
Wenlong Liu | 1 | 0 | 0.34 |
Weichao Yi | 2 | 0 | 0.34 |
Yuejin Zhao | 3 | 1 | 1.36 |
Liquan Dong | 4 | 6 | 3.10 |
Ming Liu | 5 | 13 | 2.98 |
Mei Hui | 6 | 1 | 0.69 |