非抽取小波边缘学习深度残差网络的单幅图像超分辨率重建

doi:10.12263/DZXB.20210854

电子学报 ›› 2022, Vol. 50 ›› Issue (7): 1753-1765.DOI: 10.12263/DZXB.20210854

• 学术论文 • 上一篇

非抽取小波边缘学习深度残差网络的单幅图像超分辨率重建

王相海¹^,², 赵晓阳¹, 王鑫莹², 赵克云², 宋传鸣²

^1.辽宁师范大学地理科学学院，辽宁大连 116029
^2.辽宁师范大学计算机科学与信息技术学院，辽宁大连 116081

收稿日期:2021-07-06 修回日期:2022-03-29 出版日期:2022-07-25 发布日期:2022-07-30
作者简介:王相海男，1965年11月出生于吉林省汪清县.现为辽宁师范大学地理科学学院和计算机与信息技术学院教授、博士生导师.主要研究方向为多媒体信息处理、遥感影像信息处理.E-mail： xhwang@lnnu.edu.cn
赵晓阳女，1994年4月出生于辽宁省沈阳市. 现为辽宁师范大学地理科学学院博士研究生. 主要研究方向为深度学习与遥感影像智能信息处理.E-mail: zxy_lnnu@163.com
王鑫莹女，1995年10月出生于黑龙江省齐齐哈尔市. 现为辽宁师范大学计算机与信息技术学院硕士研究生. 主要研究方向为遥感图像处理.E-mail: 1603331385@qq.com
赵克云男，1997年11月出生于辽宁省沈阳市. 现为辽宁师范大学计算机与信息技术学院硕士研究生. 主要研究方向为深度学习与遥感图像处理.E-mail: 1184234775@qq.com
宋传鸣男，1980年10月出生于辽宁省沈阳市. 现为辽宁师范大学计算机与信息技术学院教授. 主要研究方向为多媒体信息处理和智能图像处理.E-mail: chmsong@163.com
基金资助:
国家自然科学基金(41971388);辽宁省高等学校创新团队支持计划(LT2017013)

Single Image Super-Resolution Reconstruction Using Deep Residual Networks with Non-decimated Wavelet Edge Learning

WANG Xiang-hai¹^,², ZHAO Xiao-yang¹, WANG Xin-ying², ZHAO Ke-yun², SONG Chuan-ming²

^1.School of Geographical Science，Liaoning Normal University，Dalian，Liaoning 116029，China
^2.School of Computer and Information Technology，Liaoning Normal University，Dalian，Liaoning 116081，China

Received:2021-07-06 Revised:2022-03-29 Online:2022-07-25 Published:2022-07-30

摘要/Abstract

摘要：

图像超分辨率重建作为一个典型的非适定问题一直受到重视，尽管近年来出现了许多行之有效的卷积神经网络超分辨率重建模型，但如何全面挖掘图像先验信息，用以提高重建图像的细节清晰度仍有待深入研究.本文提出一种基于非抽取Wavelet变换的边缘学习深度残差网络单幅图像超分辨重建模型NDW-EDRN（Non-Decimated Wavelet Edge learning using Deep Residual Networks），在图像经非抽取Wavelet变换后获得多冗余信息、平滑及梯度值较小的低频区域和边缘及梯度值较大的高频区域的基础上，将整体网络框架设计为采用不同结构的CNN（Convolutional Neural Networks）模型来对低频子带与高频子带分别进行学习的策略：对低频子带采用稠密跳跃连接的方式整体性学习低频子带间的映射关系；对高频子带采用一种新型的U-net模型，将图像退化过程中所丢失的边缘作为网络的期望输出，通过基于块的跳跃连接来使网络更精细地学习缺失性边缘，从而更加充分、有效地获取图像在退化过程中所丢失的边缘细节信息.大量实验结果表明，该网络模型能够有效提高重建图像的质量，特别在恢复低分辨率图像的边缘信息方面具有一定的优势，在一定程度上弥补了传统CNN网络模型捕捉图像细节信息的不足.

关键词: 卷积神经网络, 残差学习, 非抽取小波变换, 图像超分辨率重建, 纹理边缘信息

Abstract:

As a typical ill-posed problem, image super-resolution reconstruction has always been paid attention to. Although many effective super-resolution reconstruction models using convolutional neural networks have been developed in recent years, how to mine the prior information in the image entirely to improve the details in reconstructed image needs to be further studied. In this paper, a single image super-resolution reconstruction model using deep residual networks with non-decimated wavelet transform edge learning NDW-EDRN(Non-Decimated Wavelet Edge learning using Deep Residual Networks)is proposed. On the basis of low-frequency regions with multiple redundant information, smoothness and low gradient values, as well as high-frequency regions with edges and high gradient values that are obtained from an image after a non-decimated wavelet transform, CNN(Convolutional Neural Networks) model with different structures are designed to learn the low and high-frequency subbands separately in the overall network framework: dense skip connection is introduced for integrating learning the mapping between low-frequency subbands; a novel U-net model is designed, which makes the edges that lost in the process of image degradation as the expected output of the network, and block-based skip connection is designed at the same time for making the network learn the lost edges more robust, to obtain the lost edge details during image degradation more sufficiently and comprehensively. A large number of experimental results show that the network can improve the quality of the reconstructed image effectively and especially has a certain advantage in recovering low-resolution images lost, and makes up for the deficiencies of traditional CNN models in image learning detail information to some extent.

Key words: convolutional neural networks, residual learning, non-decimated wavelet transform, image super-resolution reconstruction, texture edge information

中图分类号:

TP391

王相海, 赵晓阳, 王鑫莹, 赵克云, 宋传鸣. 非抽取小波边缘学习深度残差网络的单幅图像超分辨率重建[J]. 电子学报, 2022, 50(7): 1753-1765.

WANG Xiang-hai, ZHAO Xiao-yang, WANG Xin-ying, ZHAO Ke-yun, SONG Chuan-ming. Single Image Super-Resolution Reconstruction Using Deep Residual Networks with Non-decimated Wavelet Edge Learning[J]. Acta Electronica Sinica, 2022, 50(7): 1753-1765.

图/表 15

图1 图像退化过程示意图

图2 NDWT过程流程图[22]

图3 图像NDWT子带、差分子带及其直方图统计

	低频	高频
高分子带
高分直方图
	$I o r i c A$	$I o r i c H$	$I o r i c V$	$I o r i c D$
低分子带
低分直方图
	$I b i c c A$	$I b i c c H$	$I b i c c V$	$I b i c c D$
差分子带
差分直方图
	$I c A_d i f f$	$I c H_d i f f$	$I c V_d i f f$	$I c D_d i f f$
(b)　butterfly图像

图3 图像NDWT子带、差分子带及其直方图统计

	低频	高频
高分子带
高分直方图
	$I o r i c A$	$I o r i c H$	$I o r i c V$	$I o r i c D$
低分子带
低分直方图
	$I b i c c A$	$I b i c c H$	$I b i c c V$	$I b i c c D$
差分子带
差分直方图
	$I c A_d i f f$	$I c H_d i f f$	$I c V_d i f f$	$I c D_d i f f$
(b)　butterfly图像

图4 深度残差网络模型NDW-EDRN结构图

图5 块级级联示意图

表1 重建图像的定量比较1

数据集	倍数	非多尺度深度学习模型(层数≤20)								多尺度深度学习模型				本文模型
		SRCNN		VDSR		SRMD		DSRN		DWSR		WaveResNet		NDW-EDRN
		PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM
Set5	$× 2$	36.66	0.954	37.53	0.959	37.53	0.959	37.66	0.959	37.43	0.957	37.57	0.959	37.77	0.960
Set5	$× 4$	30.04	0.861	31.35	0.884	31.59	0.887	31.40	0.883	31.39	0.883	31.52	0.886	31.72	0.889
Set14	$× 2$	32.42	0.906	33.03	0.912	33.12	0.914	33.15	0.913	33.07	0.917	33.09	0.913	33.26	0.914
Set14	$× 4$	27.49	0.750	28.01	0.767	28.15	0.772	28.07	0.770	28.04	0.767	28.11	0.770	28.27	0.772
Urban100	$× 2$	29.50	0.895	30.76	0.914	30.89	0.916	30.97	0.916	30.46	0.916	30.96	0.917	31.30	0.926
Urban100	$× 4$	24.52	0.722	25.18	0.752	25.34	0.761	25.08	0.747	25.26	0.755	25.36	0.761	25.51	0.763
Average	$× 2$	32.86	0.918	33.77	0.928	33.85	0.930	33.93	0.929	33.65	0.928	33.87	0.930	34.11	0.933
Average	$× 4$	27.50	0.778	28.18	0.801	28.36	0.807	28.18	0.800	28.23	0.802	28.33	0.806	28.50	0.808

表1 重建图像的定量比较1

数据集	倍数	非多尺度深度学习模型(层数≤20)								多尺度深度学习模型				本文模型
		SRCNN		VDSR		SRMD		DSRN		DWSR		WaveResNet		NDW-EDRN
		PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM
Set5	$× 2$	36.66	0.954	37.53	0.959	37.53	0.959	37.66	0.959	37.43	0.957	37.57	0.959	37.77	0.960
Set5	$× 4$	30.04	0.861	31.35	0.884	31.59	0.887	31.40	0.883	31.39	0.883	31.52	0.886	31.72	0.889
Set14	$× 2$	32.42	0.906	33.03	0.912	33.12	0.914	33.15	0.913	33.07	0.917	33.09	0.913	33.26	0.914
Set14	$× 4$	27.49	0.750	28.01	0.767	28.15	0.772	28.07	0.770	28.04	0.767	28.11	0.770	28.27	0.772
Urban100	$× 2$	29.50	0.895	30.76	0.914	30.89	0.916	30.97	0.916	30.46	0.916	30.96	0.917	31.30	0.926
Urban100	$× 4$	24.52	0.722	25.18	0.752	25.34	0.761	25.08	0.747	25.26	0.755	25.36	0.761	25.51	0.763
Average	$× 2$	32.86	0.918	33.77	0.928	33.85	0.930	33.93	0.929	33.65	0.928	33.87	0.930	34.11	0.933
Average	$× 4$	27.50	0.778	28.18	0.801	28.36	0.807	28.18	0.800	28.23	0.802	28.33	0.806	28.50	0.808

表2 重建图像的定量比较2

数据集	倍数	生成对抗网络模型				非多尺度深度学习模型(层数＞50)						本文模型
		SRGAN		SRPGAN		DRRN_B1U25		LapSRNss-D5R8		MemNet		NDW-EDRN
		PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM
Set5	$× 2$	—	—	29.67	0.950	37.74	0.959	37.72	0.960	37.78	0.960	37.77	0.960
Set5	$× 4$	29.40	0.847	22.68	0.880	31.68	0.889	31.74	0.888	31.74	0.889	31.72	0.889
Set14	$× 2$	—	—	27.66	0.911	33.23	0.914	33.24	0.914	33.28	0.914	33.26	0.914
Set14	$× 4$	26.02	0.740	22.50	0.786	28.21	0.772	28.25	0.773	28.26	0.772	28.27	0.772
Urban100	$× 2$	—	—	30.41	0.892	31.23	0.919	31.01	0.917	31.31	0.920	31.30	0.926
Urban100	$× 4$	—	—	20.00	0.763	25.44	0.764	25.45	0.765	25.50	0.763	25.51	0.763
Average	$× 2$	—	—	20.91	0.918	34.07	0.931	33.67	0.927	34.12	0.931	34.11	0.933
Average	$× 4$	—	—	21.73	0.810	28.44	0.808	28.31	0.804	28.50	0.808	28.50	0.808

表2 重建图像的定量比较2

数据集	倍数	生成对抗网络模型				非多尺度深度学习模型(层数＞50)						本文模型
		SRGAN		SRPGAN		DRRN_B1U25		LapSRNss-D5R8		MemNet		NDW-EDRN
		PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM
Set5	$× 2$	—	—	29.67	0.950	37.74	0.959	37.72	0.960	37.78	0.960	37.77	0.960
Set5	$× 4$	29.40	0.847	22.68	0.880	31.68	0.889	31.74	0.888	31.74	0.889	31.72	0.889
Set14	$× 2$	—	—	27.66	0.911	33.23	0.914	33.24	0.914	33.28	0.914	33.26	0.914
Set14	$× 4$	26.02	0.740	22.50	0.786	28.21	0.772	28.25	0.773	28.26	0.772	28.27	0.772
Urban100	$× 2$	—	—	30.41	0.892	31.23	0.919	31.01	0.917	31.31	0.920	31.30	0.926
Urban100	$× 4$	—	—	20.00	0.763	25.44	0.764	25.45	0.765	25.50	0.763	25.51	0.763
Average	$× 2$	—	—	20.91	0.918	34.07	0.931	33.67	0.927	34.12	0.931	34.11	0.933
Average	$× 4$	—	—	21.73	0.810	28.44	0.808	28.31	0.804	28.50	0.808	28.50	0.808

图6 对比其他非多尺度网络模型放大2倍的细节图

测试图像	GT	Bicubic	SRCNN	LapSRN	VDSR	NDW-EDRN

	PSNR:	35.65	37.06	37.23	37.25	37.38
Set5中baby图像	SSIM:	0.979 8	0.985 1	0.985 6	0.985 6	0.9860

	PSNR:	22.55	22.79	23.32	23.42	23.47
Set14中baboon图像	SSIM:	0.794 2	0.827 9	0.839 1	0.840 7	0.8423

	PSNR:	21.14	24.19	23.80	24.50	24.61
Urban100中img_063图像	SSIM:	0.874 0	0.913 4	0.913 0	0.916 8	0.917 8

图7 对比非多尺度其他网络模型放大4倍的细节

测试图像	GT	Bicubic	SRCNN	LapSRN	VDSR	NDW-EDRN

	PSNR:	20.90	23.95	25.52	25.56	25.89
Set5中butterfly图像	SSIM:	0.861 4	0.925 1	0.947 0	0.946 6	0.9500

	PSNR:	28.36	29.50	29.88	29.80	29.97
Set14中lenna图像	SSIM:	0.963 2	0.970 0	0.972 2	0.971 8	0.9726

	PSNR:	21.64	23.57	24.64	24.57	25.22
Urban100中img_087图像	SSIM:	0.838 0	0.884 5	0.907 2	0.905 7	0.9161

图8 对比其他多尺度网络模型放大2倍的细节图

测试图像	GT	DWSR	WaveResNet	NDW-EDRN

	PSNR:	37.00	37.35	37.38
Set5中baby图像	SSIM:	0.984 0	0.986 1	0.986 0

	PSNR:	23.38	23.44	23.47
Set14中baboon图像	SSIM:	0.841 1	0.842 3	0.842 3

	PSNR:	24.57	24.60	24.61
Urban100中img_063图像	SSIM:	0.917 2	0.917 5	0.917 8

图9 对比其他多尺度网络模型放大4倍的细节(注：WaveResNet网络未公开四倍的预训练模型)

测试图像	GT	DWSR	WaveResNet	NDW-EDRN

	PSNR:	25.36	—	25.89
Set5中butterfly图像	SSIM:	0.944 4	—	0.950 0

	PSNR:	29.89	—	29.97
Set14中lenna图像	SSIM:	0.972 0	—	0.972 6

	PSNR:	24.85	—	25.22
Urban100中img_087图像	SSIM:	0.907 6	—	0.916 1

表3 参数与计算量比较

SR_

CNN

VD_

SR_

DR_

Lap_

SRN

Mem_

Net

NSW

-EDRN

表4 消融实验重建数据集的定量比较

数据集	倍数	学习模式
		整体式		缺失式		NDW-EDRN
		PSNR	SSIM	PSNR	SSIM	PSNR	SSIM
Set5	×2	37.69	0.952	37.74	0.957	37.77	0.960
Set14		33.14	0.909	33.23	0.912	33.26	0.914
Urban 100		31.22	0.921	31.26	0.925	31.30	0.926
Aver age	×2	34.02	0.927	34.08	0.931	34.11	0.933

图10 消融实验放大2倍的细节(左上角:GT,右上角:“整体式”学习,左下角:“缺失式_s”学习,右下角:NDW-EDRN_s)

表5 消融实验重建图像的定量比较

测试图像	学习模式
	整体式		缺失式_s		NDW-EDRN_s
	PSNR	SSIM	PSNR	SSIM	PSNR	SSIM
bird	38.39	0.994 7	38.42	0.995 1	38.46	0.995 3
butter fly	31.95	0.987 4	32.00	0.987 6	32.02	0.987 5
img 068	35.79	0.974 1	35.82	0.974 5	35.88	0.974 8
baby	36.61	0.984 5	36.58	0.984 4	36.64	0.984 6

参考文献 39

1	LECUN Y, BOSER B E, DENKER J S, et al. Backpropagation applied to handwritten zip code recognition[J]. Neural Computation, 1989, 1(4): 541-551.
2	LECUN Y, BOTTOU L. Gradient-based learning applied to document recognition[J]. Proceedings of the IEEE, 1998, 86(11): 2278-2324.
3	周飞燕,金林鹏,董军.卷积神经网络研究综述[J].计算机学报, 2017, 40(6): 1229-1251.
	Zhou Fei-yan, Jin Lin-peng, Dong Jun. Review of convolutional neural network[J]. Chinese Journal of Computers, 2017, 40(6): 1229-1251. (in Chinese)
4	KHAN S, RAHMANI H, SHAH S A A, et al. A guide to convolutional neural networks for computer vision[J]. Synthesis Lectures on Computer Vision, 2018, 8(1): 1-207.
5	DONG C, LOY C C, HE K, TANG X. Learning a deep convolutional network for image super-resolution[C]//Proceedings of the 13th European Conference on Computer Vision. Berlin: Springer, 2014: 184-199.
6	DONG C, LOY C C, TANG X. Accelerating the super-resolution convolutional neural network[C]//Proceedings of the European Conference on Computer Vision. Berlin: Springer, 2016: 391-407.
7	KIM J, LEE J K, LEE K M. Accurate image super-resolution using very deep convolutional networks[C]//Proceedings of the Computer Vision and Pattern Recognition. Las Vegas: IEEE, 2016: 1646-1654.
8	KIM J, LEE J K, LEE K M. Deeply-recursive convolutional network for image super-resolution[C]//Proceedings of the Computer Vision and Pattern Recognition. Piscataway, NJ: IEEE, 2016: 1637-1645.
9	SHI W Z, CABALLERO J, HUSZAR F, et al. Real-time single image and video super-resolution using an efficient sub-pixel convolutional neural network[C]//Proceedings of the Computer Vision and Pattern Recognition. Piscataway, NJ: IEEE, 2016: 1874-1883.
10	LAI W S, HUANG J B, AHUJA N, et al. Fast and accurate image super-resolution with deep laplacian pyramid networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 41(11): 2599-2613.
11	CHRISTIAN L, LUCAS T, FERENC H, et al. Photo-realistic single image super-resolution using a generative adversarial network[C]//Proceedings of the IEEE conference on Computer Vision and Pattern Recognition. Honolulu: IEEE, 2017: 105-114.
12	WU B, DUAN H, LIU Z, et al. SRPGAN: Perceptual generative adversarial network for single image super resolution[J/OL]. (2017-12-20) [2021-06-20]. .
13	WANG X, YU K, WU S, et al. Esrgan: Enhanced super-resolution generative adversarial networks[C]//Proceedings of the European Conference on Computer Vision. Munich: Springer, 2018: 63-79.
14	JOLICOEUR-MARTINEAU A. The relativistic discriminator: A key element missing from standard GAN[J/OL]. (2018-09-10)[2021-06-25]. .
15	宋传鸣, 赵长伟, 刘丹, 王相海. 3D多尺度几何分析研究进展[J].软件学报, 2015, 26(5): 1213-1236.
	Song Chuan-ming, Zhao Chang-wei, Liu Dan, Wang Xiang-hai. Advances in three-dimensional multiscale geometrical analysis[J]. Journal of Software, 2015, 26(5): 1213-1236. (in Chinese)
16	VYAS A, YU S, PAIK J. Multiscale Transforms with Application to Image Processing: Wavelets and Wavelet transform[M]. Singapore: Springer, 2018.
17	BAE W, YOO J, YE J C. Beyond deep residual learning for image restoration: Persistent homology-guided manifold simplification[C]//Proceedings of the Computer Vision and Pattern Recognition Workshops. Honolulu: IEEE, 2017:1141-1149.
18	GUO T, MOUSAVI H S, VU T H, et al. Deep wavelet prediction for image super-resolution[C]//Proceedings of the Computer Vision and Pattern Recognition Workshops. Honolulu: IEEE, 2017: 1100-1109.
19	YUE L, SHEN H, LI J, et al. Image super-resolution: The techniques, applications, and future[J]. Signal processing, 2016, 128(11):389-408.
20	SHENSA M J. The discrete wavelet transform: Wedding the atrous and mallat algorithms[J]. IEEE Transactions on Signal Processing, 1992, 40(10): 2464-2482.
21	NASON G P, SILVERMAN B W. The stationary wavelet transform and some statistical applications[C]//Wavelets and Statistics Lecture Notes in Statistics. Berlin: Springer,1995: 281-300.
22	NASON G P. Wavelet methods in statistics with R[J]. Journal of the Royal Statistical Society, 2010, 173(1): 273-273.
23	DAUBECHIES I. Orthonormal bases of compactly supported wavelets[J]. Comms Pure Appl Math, 1988, 41(7): 909-996.
24	BEVILACQUA M, ROUMY A, GUILLEMOT C, et al. Low-complexity single image super-resolution based on nonnegative neighbor embedding[C]//Proceedings of the British Machine Vision Conference. Guildford, UK: BMVA Press, 2012: 135-144.
25	XIN D, REN Y, MAI X, et al. Wavelet domain style transfer for an effective perception-distortion tradeoff in single image super-resolution[C]//Proceedings of the IEEE International Conference on Computer Vision. Seoul: IEEE, 2019: 3076-3085.
26	TONG T, LI G, LIU X, et al. Image super-resolution using dense skip connections[C]//Proceedings of the IEEE International Conference on Computer Vision. Venice: IEEE, 2017: 4809-4817.
27	TONG T, LI G, LIU X, et al. Image super-resolution using dense skip connections[C]//Proceedings of the IEEE International Conference on Computer Vision. Piscataway, NJ: IEEE, 2017: 4809-4817.
28	RONNEBERGER O, FISCHER P, BROX T. U-Net: Convolutional networks for biomedical image segmentation[C]//Proceedings of the International Conference on Medical Image Computing & Computer-assisted Intervention. Munich: Springer, 2015: 234-241.
29	RONNEBERGER O, FISCHER P, BROX T. U-net: Convolutional networks for biomedical image segmentation[C]//Proceedings of the Medical Image Computing and Computer-Assisted Intervention. Berlin: Springer, 2015: 234-241.
30	HE K, ZHANG X, REN S, et al. Deep residual learning for image recognition[C]//Proceedings of the Computer Vision and Pattern Recognition. Las Vegas: IEEE, 2016: 770-778.
31	ARBELAEZ P, MAIRE M, FOWLKES C, et al. Contour detection and hierarchical image segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(5): 898-916.
32	YAGN J, WEIGHT J, HUANG T S, MA Y. Image super-resolution via sparse representation[J]. IEEE Transactions Image Processing, 2010, 19(11): 2861-2873.
33	ZEYDE R, ELAD M, PROTTER M. On single image scale-up using sparse-representations[C]//Proceedings of the 7th International Conference Curves Surfaces. Berlin: Springer,2012: 11-730.
34	HUANG J B, SINGH A, AHUJA N. Single image super-resolution from transformed self-exemplars[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. Boston: IEEE, 2015: 5197-5206.
35	KINGMA D, BA J. Adam: A method for stochastic optimization[J/OL]. (2017-01-17)[2021-05-20]. .
36	ZHANG K, ZUO W, ZHANG L. Learning a single convolutional super-resolution network for multiple degradations[C]//Proceedings of the IEEE Conference on Computer Vision & Pattern Recognition. Salt Lake City: IEEE, 2018: 3262-3271.
37	HAN W, CHANG S, LIU D, et al. Image super-resolution via dual-state recurrent networks[C]//Proceedings of the IEEE Conference on Computer Vision & Pattern Recognition. Salt Lake City: IEEE, 2018: 1654-1663.
38	TAI Y, YANG J, LIU X. Image super-resolution via deep recursive residual network[C]//Proceedings of the IEEE Conference on Computer Vision & Pattern Recognition. Honolulu: IEEE, 2017: 2790-2798.
39	TAI Y, YANG J, LIU X, XU C. Memnet: A persistent memory network for image restoration[C]//Proceedings of the IEEE International Conference on Computer Vision. Venice: IEEE, 2017:4549-4557.

[1]	张志文, 刘天歌, 聂鹏举. 基于实景数据增强和双路径融合网络的实时街景语义分割算法[J]. 电子学报, 2022, 50(7): 1609-1620.
[2]	丁毅, 沈薇, 李海生, 钟琼慧, 田明宇, 李洁. 面向CNN的区块链可信隐私服务计算模型[J]. 电子学报, 2022, 50(6): 1399-1409.
[3]	张波, 陆云杰, 秦东明, 邹国建. 一种卷积自编码深度学习的空气污染多站点联合预测模型[J]. 电子学报, 2022, 50(6): 1410-1427.
[4]	廖勇, 李玉杰. 一种轻量化低复杂度的FDD大规模MIMO系统CSI反馈方法[J]. 电子学报, 2022, 50(5): 1211-1217.
[5]	张云, 化青龙, 姜义成, 徐丹. 基于混合型复数域卷积神经网络的三维转动舰船目标识别[J]. 电子学报, 2022, 50(5): 1042-1049.
[6]	崔亚奇, 何友, 唐田田, 熊伟. 一种深度学习航迹关联方法[J]. 电子学报, 2022, 50(3): 759-763.
[7]	伍邦谷, 张苏林, 石红, 朱鹏飞, 王旗龙, 胡清华. 基于多分支结构的不确定性局部通道注意力机制[J]. 电子学报, 2022, 50(2): 374-382.
[8]	李居朋, 王颖慧, 李刚. 医学图像关键点检测深度学习方法研究与挑战[J]. 电子学报, 2022, 50(1): 226-237.
[9]	赵琰, 赵凌君, 匡纲要. 基于注意力机制特征融合网络的SAR图像飞机目标快速检测[J]. 电子学报, 2021, 49(9): 1665-1674.
[10]	管永明, 王刚, 骆凯波, 吕梁, 吕晓雯, 史玉良. 基于极端不平衡学习的泛化低压异常箱表关系识别研究与应用[J]. 电子学报, 2021, 49(8): 1507-1514.
[11]	徐频捷, 陈逸杰, 李之南, 赵地. 基于事件驱动的车道线识别算法研究[J]. 电子学报, 2021, 49(7): 1379-1385.
[12]	罗会兰, 袁璞, 童康. 基于深度学习的显著性目标检测方法综述[J]. 电子学报, 2021, 49(7): 1417-1427.
[13]	程旭, 宋晨, 史金钢, 周琳, 张毅锋, 郑钰辉. 基于深度学习的通用目标检测研究综述[J]. 电子学报, 2021, 49(7): 1428-1438.
[14]	张昱, 刘开峰, 张全新, 王艳歌, 高凯龙. 基于组合-卷积神经网络的中文新闻文本分类[J]. 电子学报, 2021, 49(6): 1059-1067.
[15]	裴翰奇, 杨春玲, 魏志超, 曹燕. 基于SPL迭代思想的图像压缩感知重构神经网络[J]. 电子学报, 2021, 49(6): 1195-1203.

非抽取小波边缘学习深度残差网络的单幅图像超分辨率重建

Single Image Super-Resolution Reconstruction Using Deep Residual Networks with Non-decimated Wavelet Edge Learning

RichHTML

PDF

可视化

摘要/Abstract

引用本文

使用本文

图/表 15

参考文献 39

相关文章 15

编辑推荐

Metrics