Image Super-Resolution via Attention Based Back Projection Networks

Abstract

Deep learning based image Super-Resolution (SR) has shown rapid development due to its ability of big data digestion. Generally, deeper and wider networks can extract richer feature maps and generate SR images with remarkable quality. However, the more complex network we have, the more time consumption is required for practical applications. It is important to have a simplified network for efficient image SR. In this paper, we propose an Attention based Back Projection Network (ABPN) for image superresolution. Similar to some recent works, we believe that the back projection mechanism can be further developed for SR. Enhanced back projection blocks are suggested to iteratively update low-and high-resolution feature residues. Inspired by recent studies on attention models, we propose a Spatial Attention Block (SAB) to learn the cross-correlation across features at different layers. Based on the assumption that a good SR image should be close to the original LR image after down-sampling. We propose a Refined Back Projection Block (RBPB) for final reconstruction. Extensive experiments on some public and AIM2019 Image Super-Resolution Challenge [4] datasets show that the proposed ABPN can provide state-of-the-art or even better performance in both quantitative and qualitative measurements.

Single image super‐resolution based on sparse representation using dictionaries trained with input image patches

“…In the ABPN method [15], the size of the LR image patches is 32 × 32 and the size of the SR image patches is 64 × 64. In the IKC method [16], the kernel size is 21 × 21.…”

“…Methods in [15, 16] utilise deep neural networks. In [15], the authors proposed an attention‐based back projection network (ABPN) for SR.…”

“…Methods in [15, 16] utilise deep neural networks. In [15], the authors proposed an attention‐based back projection network (ABPN) for SR. Since the shrinkage of the SR image should be close to that of the LR image, they proposed enhanced back‐projection blocks to iteratively update the LR and HR feature residues.…”

Modified Dual Path Network With Transform Domain Data for Image Super-Resolution

“…Inspired by [24] and [25], we also adopted the spatial attention block attached at each DPB to learn the correlations between hierarchical features as shown in Fig. 3.…”

“…Specifically, in FENet, we adopt a convolutional layer to extract the initial features from the concatenation of LR input images and the transformed inputs. The self-attention block proposed in [25] is located at the end of FENet to recalibrate the features. Note that the self-attention takes only the current input for computation, as shown in Fig.…”