文献类型：期刊文献

英文题名：MRFN: Multi-Receptive-Field Network for Fast and Accurate Single Image Super-Resolution

作者：He, Zewei[1,2];Cao, Yanpeng[1,2];Du, Lei[3];Xu, Baobei[1,2];Yang, Jiangxin[1,2];Cao, Yanlong[1,2];Tang, Siliang[4];Zhuang, Yueting[4]

机构：[1]Zhejiang Univ, State Key Lab Fluid Power & Mechatron Syst, Sch Mechan Engn, Hangzhou 310027, Peoples R China;[2]Zhejiang Univ, Key Lab Adv Mfg Technol Zhejiang Prov, Sch Mechan Engn, Hangzhou 310027, Peoples R China;[3]Shaoxing Univ, Yuanpei Coll, Shaoxing 312000, Peoples R China;[4]Zhejiang Univ, Coll Comp Sci & Technol, Hangzhou 310027, Peoples R China

年份：2020

卷号：22

期号：4

起止页码：1042

外文期刊名：IEEE TRANSACTIONS ON MULTIMEDIA

收录：SCI-EXPANDED(收录号：WOS:000522440400018)、、EI(收录号：20201508402542)、Scopus(收录号：2-s2.0-85082883256)、WOS

基金：This work was supported by the National Natural Science Foundation of China (No. 51605428, 51575486, and 61751209.)

语种：英文

外文关键词：Single image super-resolution; deep learning; multi-receptive-field; loss function

外文摘要：Recently, convolutional neural network (CNN) based models have shown great potential in the task of single image super-resolution (SISR). However, many state-of-the-art SISR solutions are reproducing some tricks proven effective in other vision tasks, such as pursuing a deeper model. In this paper, we propose a new solution (named as Multi-Receptive-Field Network - MRFN), which outperforms existing SISR solutions in three different aspects. First, from receptive field: a novel multi-receptive-field (MRF) module is proposed to extract and fuse features in different receptive fields from local to global. Integrating these hierarchical features can generate better mappings on recovering high-fidelity details at different scales. Second, from network architectures: both dense skip connections and deep supervision are utilized to combine features from the current MRF module and preceding ones for training more representative features. Moreover, a deconvolution layer is embedded at the end of the network to avoid artificial priors induced by numerical data pre-processing (e.g., bicubic stretching), and speed up the restoration process. Finally, from error modeling: different from L1 and L2 loss functions, we proposed a novel two-parameter training loss called Weighted Huber loss function which can adaptively adjust the value of back-propagated derivative according to the residual value, thus fit the reconstruction error more effectively. Extensive qualitative and quantitative evaluation results on benchmark datasets demonstrate that our proposed MRFN can achieve more accurate recovering results than most state-of-the-art methods with significantly less complexity.