Video Super-Resolution via Deep Draft-Ensemble Learning
Authors:
- Renjie LiaoRLRenjie Liao
- Xin TaoXTXin Tao
- Ruiyu LiRLRuiyu Li
- Ziyang MaZMZiyang Ma
- Jiaya JiaJJJiaya Jia
- 1 December 2015
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 531-539
Abstract
We propose a new direction for fast video super-resolution (VideoSR) via a SR draft ensemble, which is defined as the set of high-resolution patch candidates before final image deconvolution. Our method contains two main components -- i.e., SR draft ensemble generation and its optimal reconstruction. The first component is to renovate traditional feedforward reconstruction pipeline and greatly enhance its ability to compute different super resolution results considering large motion variation and possible errors arising in this process. Then we combine SR drafts through the nonlinear process in a deep convolutional neural network (CNN). We analyze why this framework is proposed and explain its unique advantages compared to previous iterative methods to update different modules in passes. Promising experimental results are shown on natural video sequences.Keywords
This publication has 12 references indexed in Scilit:
- Caffe2014
- Restoring an Image Taken through a Window Covered with Dirt or Rain2013
- Super-Resolution Without Explicit Subpixel Motion EstimationIEEE Transactions on Image Processing, 2009
- Generalizing the Nonlocal-Means to Super-Resolution ReconstructionIEEE Transactions on Image Processing, 2008
- Fast image/video upsamplingACM Transactions on Graphics, 2008
- Fast and Robust Multiframe Super ResolutionIEEE Transactions on Image Processing, 2004
- A fast super-resolution reconstruction algorithm for pure translational motion and common space-invariant blurIEEE Transactions on Image Processing, 2001
- Improved resolution from subpixel shifted picturesCVGIP: Graphical Models and Image Processing, 1992
- Improving resolution by image registrationCVGIP: Graphical Models and Image Processing, 1991
- Recursive reconstruction of high resolution image from noisy undersampled multiframesIEEE Transactions on Acoustics, Speech, and Signal Processing, 1990