Image super-resolution via enhanced multi-scale residual network

Abstract

Recently, a very deep convolutional neural network (CNN) has achieved impressive results in image super-resolution (SR). In particular, residual learning techniques are widely used. However, the previously proposed residual block can only extract one single-level semantic feature maps of one single receptive field. Therefore, it is necessary to stack the residual blocks to extract higher-level semantic feature maps, which will significantly deepen the network. While a very deep network is hard to train and limits the representation for reconstructing the hierarchical information. Based on the residual block, we propose an enhanced multi-scale residual network (EMRN) to take advantage of hierarchical image features via dense connected enhanced multi-scale residual blocks (EMRBs). Specifically, the newly proposed residual block (EMRB) is capable of constructing multi-level semantic feature maps by a two-branch inception. The twobranch inception in our proposed EMRB consists of 2 convolutional layers and 4 convolutional layers in each branch respectively, therefore we have different ranges of receptive fields within one single EMRB. Meanwhile, the local feature fusion (LFF) is used in every EMRB to adaptively fuse the local feature maps extracted by the two-branch inception. Furthermore, global feature fusion (GFF) in EMRN is then used to obtain abundant useful features from previous EMRBs and subsequent ones in a holistic manner. Experiments on benchmark datasets suggest that our EMRN performs favorably over the state-of-the-art methods in reconstructing further superior super-resolution (SR) images.

Single‐image super‐resolution using lightweight transformer‐convolutional neural network hybrid model

“…To prove the effectiveness and to advance the method in this paper, we selected recent deep learning methods for comparison with our method, including SRCNN [9], VDSR [11], EDSR [28], fast, accurate, and lightweight super‐resolution with cascading residual network (CARN) [52], IDN [53], RCAN [29], RDN [37], RFDN [54], SMSR [55], NLSN [39], BSRN [40], DRMSFFN [56], and EMRN [57]. The comparisons were made at two, three, and four times the scale, and the results are shown in Table 1.…”

Dense Convolutional Network and Its Application in Medical Image Analysis

“…Yang et al [ 43 ] introduced convolutional residual in DenseNet by parallelizing multiple multiresolution convolutional residuals in dense block, effectively reducing the number of superresolution network parameters, and increasing model hierarchy and network depth, with prediction speed reaches 25 ms per image under certain accuracy loss. Single-branch convolutional residual block can only extract single-level semantic information; Wang et al [ 44 ] designed multiscale residual network (EMRN) in image superresolution task and densely connected multiscale residual blocks extracted image hierarchical features to achieve multilevel semantic information in different perceptual domains.…”

Cascaded Residual Densely Connected Network for Image Super-Resolution

“…In order to deal with special domain application such as extreme motion blur and recover sharp high-resolution (HR) image [22], thermal image [23][24], X-ray image [25], 3D image SR [26], Stereo Image [27], Remote Sensing Image [18] and so on. Multi-level image SR, such as Pyramidal structure [28], Multi-Scale SR [29]- [31], GAN(Generative neural network) [32], RNN(Recurrent neural network) [33], Cascade [17][18] and so on.…”

“…Among the current methods, residual learning and dense connection are the most import structure. Residual learning [26][35]- [37] is constructed between the input low resolution image and the final output high-resolution images. This connection can also be called global connection.…”