Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Hierarchical Similarity Learning for Aliasing Suppression Image Super-Resolution

Yuqing Liu, Qi Jia, Jian Zhang, Xin Fan, Shanshe Wang, Siwei Ma, Wen Gao

TL;DR

The self-similarity is considered and a hierarchical image super-resolution network (HSRNet) is proposed to suppress the influence of aliasing and the experimental result shows that HSRNet achieves better quantitative and visual performance than other works and remits the aliasing more effectively.

Abstract

As a highly ill-posed issue, single image super-resolution (SISR) has been widely investigated in recent years. The main task of SISR is to recover the information loss caused by the degradation procedure. According to the Nyquist sampling theory, the degradation leads to aliasing effect and makes it hard to restore the correct textures from low-resolution (LR) images. In practice, there are correlations and self-similarities among the adjacent patches in the natural images. This paper considers the self-similarity and proposes a hierarchical image super-resolution network (HSRNet) to suppress the influence of aliasing. We consider the SISR issue in the optimization perspective, and propose an iterative solution pattern based on the half-quadratic splitting (HQS) method. To explore the texture with local image prior, we design a hierarchical exploration block (HEB) and progressive increase the receptive field. Furthermore, multi-level spatial attention (MSA) is devised to obtain the relations of adjacent feature and enhance the high-frequency information, which acts as a crucial role for visual experience. Experimental result shows HSRNet achieves better quantitative and visual performance than other works, and remits the aliasing more effectively.

Hierarchical Similarity Learning for Aliasing Suppression Image Super-Resolution

TL;DR

The self-similarity is considered and a hierarchical image super-resolution network (HSRNet) is proposed to suppress the influence of aliasing and the experimental result shows that HSRNet achieves better quantitative and visual performance than other works and remits the aliasing more effectively.

Abstract

As a highly ill-posed issue, single image super-resolution (SISR) has been widely investigated in recent years. The main task of SISR is to recover the information loss caused by the degradation procedure. According to the Nyquist sampling theory, the degradation leads to aliasing effect and makes it hard to restore the correct textures from low-resolution (LR) images. In practice, there are correlations and self-similarities among the adjacent patches in the natural images. This paper considers the self-similarity and proposes a hierarchical image super-resolution network (HSRNet) to suppress the influence of aliasing. We consider the SISR issue in the optimization perspective, and propose an iterative solution pattern based on the half-quadratic splitting (HQS) method. To explore the texture with local image prior, we design a hierarchical exploration block (HEB) and progressive increase the receptive field. Furthermore, multi-level spatial attention (MSA) is devised to obtain the relations of adjacent feature and enhance the high-frequency information, which acts as a crucial role for visual experience. Experimental result shows HSRNet achieves better quantitative and visual performance than other works, and remits the aliasing more effectively.
Paper Structure (18 sections, 20 equations, 14 figures, 7 tables)

This paper contains 18 sections, 20 equations, 14 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Literature Review
  3. Deep learning for Single Image Super-Resolution
  4. Image Super-Resolution in the Optimization View
  5. Methodology
  6. Network Design
  7. Hierarchical Exploration Block
  8. Multi-level Spatial Attention
  9. Experiment
  10. Settings
  11. Model Analysis
  12. Investigation on the Optimization Modules
  13. Investigation on HEB
  14. Investigation on MSA
  15. Investigation on Iterative Design
  16. ...and 3 more sections

Figures (14)

  • Figure 1: Visual quality comparison for different SISR methods.
  • Figure 2: Network structure of HSRNet. The space transformation is conducted by convolution and upscale. Three components are proposed to perform the optimization
  • Figure 3: Structure of hierarchical exploration block (HEB).
  • Figure 4: Comparison between the hierarchical exploration in HEB and the Laplacian Pyramid.
  • Figure 5: Structure of multi-level spatial attention (MSA).
  • ...and 9 more figures