High-Similarity-Pass Attention for Single Image Super-Resolution

TL;DR

This paper introduces high-similarity-pass attention (HSPA), a novel attention mechanism for single image super-resolution that improves efficiency and interpretability by filtering irrelevant features, leading to superior reconstruction performance.

Contribution

The paper proposes HSPA with a soft thresholding operation, providing a more compact and interpretable attention distribution, and demonstrates its integration into deep SISR models for enhanced results.

Findings

HSPAN outperforms state-of-the-art methods in quantitative metrics.

HSPA produces more compact and interpretable attention maps.

The approach effectively models long-range dependencies with reduced redundancy.

Abstract

Recent developments in the field of non-local attention (NLA) have led to a renewed interest in self-similarity-based single image super-resolution (SISR). Researchers usually used the NLA to explore non-local self-similarity (NSS) in SISR and achieve satisfactory reconstruction results. However, a surprising phenomenon that the reconstruction performance of the standard NLA is similar to the NLA with randomly selected regions stimulated our interest to revisit NLA. In this paper, we first analyzed the attention map of the standard NLA from different perspectives and discovered that the resulting probability distribution always has full support for every local feature, which implies a statistical waste of assigning values to irrelevant non-local features, especially for SISR which needs to model long-range dependence with a large number of redundant non-local features. Based on these findings, we introduced a concise yet effective soft thresholding operation to obtain high-similarity-pass attention (HSPA), which is beneficial for generating a more compact and interpretable distribution. Furthermore, we derived some key properties of the soft thresholding operation that enable training our HSPA in an end-to-end manner. The HSPA can be integrated into existing deep SISR models as an efficient general building block. In addition, to demonstrate the effectiveness of the HSPA, we constructed a deep high-similarity-pass attention network (HSPAN) by integrating a few HSPAs in a simple backbone. Extensive experimental results demonstrate that HSPAN outperforms state-of-the-art approaches on both quantitative and qualitative evaluations.