Dynamically Unfolding Recurrent Restorer: A Moving Endpoint Control Method for Image Restoration

  • 2018-10-10 16:12:37
  • Xiaoshuai Zhang, Yiping Lu, Jiaying Liu, Bin Dong
  • 1

Abstract

In this paper, we propose a new control framework called the moving endpointcontrol to restore images corrupted by different degradation levels in onemodel. The proposed control problem contains a restoration dynamics which ismodeled by an RNN. The moving endpoint, which is essentially the terminal timeof the associated dynamics, is determined by a policy network. We call theproposed model the dynamically unfolding recurrent restorer (DURR). Numericalexperiments show that DURR is able to achieve state-of-the-art performances onblind image denoising and JPEG image deblocking. Furthermore, DURR can wellgeneralize to images with higher degradation levels that are not included inthe training stage.

 

Introduction (beta)

None

 

Conclusion (beta)

None