FormNet : Formatted Learning for Image Restoration

Research output: Journal Publications and Reviews (RGC: 21, 22, 62)21_Publication in refereed journal

1 Scopus Citations
View graph of relations

Author(s)

  • Jianbo Jiao
  • Wei-Chih Tu
  • Ding Liu
  • Shengfeng He
  • Thomas S. Huang

Related Research Unit(s)

Detail(s)

Original languageEnglish
Pages (from-to)6302-6314
Journal / PublicationIEEE Transactions on Image Processing
Volume29
Online published6 May 2020
Publication statusPublished - 2020

Abstract

In this paper, we propose a deep CNN to tackle the image restoration problem by learning formatted information. Previous deep learning based methods directly learn the mapping from corrupted images to clean images, and may suffer from the gradient exploding/vanishing problems of deep neural networks. We propose to address the image restoration problem by learning the structured details and recovering the latent clean image together, from the shared information between the corrupted image and the latent image. In addition, instead of learning the pure difference (corruption), we propose to add a residual formatting layer and an adversarial block to format the information to structured one, which allows the network to converge faster and boosts the performance. Furthermore, we propose a cross-level loss net to ensure both pixel-level accuracy and semantic-level visual quality. Evaluations on public datasets show that the proposed method performs favorably against existing approaches quantitatively and qualitatively.

Research Area(s)

  • Image restoration, format, residual, GAN, CNN

Citation Format(s)

FormNet : Formatted Learning for Image Restoration. / Jiao, Jianbo; Tu, Wei-Chih; Liu, Ding; He, Shengfeng; Lau, Rynson W. H.; Huang, Thomas S.

In: IEEE Transactions on Image Processing, Vol. 29, 2020, p. 6302-6314.

Research output: Journal Publications and Reviews (RGC: 21, 22, 62)21_Publication in refereed journal