《计算机应用》唯一官方网站 ›› 2023, Vol. 43 ›› Issue (7): 2280-2287.DOI: 10.11772/j.issn.1001-9081.2022060877
所属专题: 多媒体计算与计算机仿真
收稿日期:
2022-06-16
修回日期:
2022-09-06
接受日期:
2022-09-08
发布日期:
2022-10-18
出版日期:
2023-07-10
通讯作者:
梁敏
作者简介:
梁敏(1979—),女,山西忻州人,副教授,博士,CCF会员,主要研究方向:图像处理、模式识别;基金资助:
Min LIANG(), Jiayi LIU, Jie LI
Received:
2022-06-16
Revised:
2022-09-06
Accepted:
2022-09-08
Online:
2022-10-18
Published:
2023-07-10
Contact:
Min LIANG
About author:
LIANG Min, born in 1979, Ph. D., associate professor. Her research interests include image processing, pattern recognition.Supported by:
摘要:
针对图像超分辨重建过程中原始高清图片与低质量图像之间缺乏依赖关系、深度网络中特征图信息不分主次重构导致的图像高频信息高精度重构困难的问题,提出一种融合迭代反馈与注意力机制的单幅图像超分辨重建方法。首先使用频率分解模块分别提取图像中的高、低频信息,并将二者分别处理,使网络重点关注提取出的高频细节部分,增强方法在图像细节上的复原能力;其次通过通道注意力机制将重建的重点放在有效特征所在的特征通道上,增强网络提取特征图信息的能力;然后采用迭代反馈的思想,在反复重建和比对过程中增加图像的还原程度;最后通过重建模块生成输出图像。在Set5、Set14、BSD100、Urban100和Manga109基准数据集上的2倍、4倍和8倍放大实验中,与主流超分辨率方法相比,所提方法表现出更优越的性能。在Manga109数据集的8倍放大实验中,相较于传统插值方法和基于卷积神经网络的图像超分辨率算法(SRCNN),所提方法的峰值信噪比(PSNR)均值分别提升了约3.01 dB和2.32 dB。实验结果表明:所提方法能够降低重建过程中出现的误差,并有效重建出更精细的高分辨率图像。
中图分类号:
梁敏, 刘佳艺, 李杰. 融合迭代反馈与注意力机制的图像超分辨重建方法[J]. 计算机应用, 2023, 43(7): 2280-2287.
Min LIANG, Jiayi LIU, Jie LI. Image super-resolution reconstruction method based on iterative feedback and attention mechanism[J]. Journal of Computer Applications, 2023, 43(7): 2280-2287.
频率分解模块 | 注意力模块 | PSNR/dB |
---|---|---|
× | × | 26.80 |
× | √ | 26.89 |
√ | × | 26.85 |
√ | √ | 26.90 |
表1 不同模型结构在测试集上的实验结果(α=8)
Tab. 1 Experimental results of different model structures on test set (α=8)
频率分解模块 | 注意力模块 | PSNR/dB |
---|---|---|
× | × | 26.80 |
× | √ | 26.89 |
√ | × | 26.85 |
√ | √ | 26.90 |
λ值 | PSNR/dB | λ值 | PSNR/dB |
---|---|---|---|
0 | 26.30 | 1 | 25.12 |
0.01 | 27.08 | 10 | 23.41 |
0.1 | 27.03 |
表2 不同λ值在General100数据集上的实验结果(α=4)
Tab. 2 Experimental results for different λ values on General100 dataset (α=4)
λ值 | PSNR/dB | λ值 | PSNR/dB |
---|---|---|---|
0 | 26.30 | 1 | 25.12 |
0.01 | 27.08 | 10 | 23.41 |
0.1 | 27.03 |
算法 | 放大 倍数 | Set5 | Set14 | BSD100 | Urban100 | Manga109 | |||||
---|---|---|---|---|---|---|---|---|---|---|---|
PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | ||
Bicubic | 2× | 33.65 | 0.930 | 30.24 | 0.869 | 29.56 | 0.844 | 26.88 | 0.841 | 30.84 | 0.935 |
SRCNN[ | 36.66 | 0.954 | 32.45 | 0.906 | 31.36 | 0.888 | 29.52 | 0.895 | 35.72 | 0.968 | |
FSRCNN[ | 37.00 | 0.956 | 32.63 | 0.909 | 31.50 | 0.891 | 29.88 | 0.902 | — | — | |
VDSR[ | 37.74 | 0.959 | 32.97 | 0.913 | 31.90 | 0.896 | 30.77 | 0.914 | 37.16 | 0.974 | |
LapSRN[ | 37.52 | 0.929 | 33.08 | 0.913 | 31.80 | 0.895 | 31.05 | 0.910 | 37.53 | 0.974 | |
ADSR[ | 37.36 | 0.958 | 32.86 | 0.911 | 31.78 | 0.894 | 30.44 | 0.910 | — | — | |
DBPN[ | 37.60 | 0.959 | 33.18 | 0.914 | 31.94 | 0.897 | 31.20 | 0.918 | 37.57 | 0.974 | |
IFANet(本文方法) | 37.70 | 0.959 | 33.24 | 0.914 | 31.99 | 0.898 | 31.15 | 0.912 | 37.85 | 0.975 | |
Bicubic | 4× | 28.42 | 0.810 | 26.10 | 0.702 | 25.96 | 0.667 | 23.15 | 0.657 | 24.92 | 0.789 |
SRCNN[ | 30.49 | 0.862 | 27.61 | 0.751 | 26.91 | 0.710 | 24.53 | 0.722 | 27.66 | 0.858 | |
FSRCNN[ | 30.71 | 0.865 | 27.70 | 0.756 | 26.97 | 0.714 | 24.61 | 0.727 | 27.89 | 0.859 | |
VDSR[ | 31.53 | 0.883 | 28.03 | 0.767 | 27.29 | 0.725 | 25.18 | 0.752 | 28.82 | 0.886 | |
SRGAN[ | 29.46 | 0.838 | 26.60 | 0.718 | 25.74 | 0.666 | 24.50 | 0.736 | 27.79 | 0.856 | |
LapSRN[ | 31.54 | 0.885 | 28.09 | 0.770 | 27.31 | 0.727 | 25.21 | 0.756 | 29.09 | 0.890 | |
ADSR[ | 31.19 | 0.881 | 27.88 | 0.763 | 27.20 | 0.721 | 25.00 | 0.744 | — | — | |
DBPN[ | 31.76 | 0.887 | 28.39 | 0.778 | 27.48 | 0.733 | 25.71 | 0.772 | 30.22 | 0.902 | |
IFANet (本文方法) | 32.09 | 0.890 | 28.30 | 0.776 | 27.75 | 0.738 | 25.90 | 0.786 | 30.65 | 0.912 | |
Bicubic | 8× | 24.39 | 0.657 | 23.19 | 0.568 | 23.67 | 0.547 | 20.74 | 0.515 | 21.68 | 0.649 |
SRCNN[ | 25.33 | 0.689 | 23.85 | 0.593 | 24.13 | 0.565 | 21.29 | 0.543 | 22.37 | 0.682 | |
FSRCNN[ | 25.41 | 0.682 | 23.93 | 0.592 | 24.21 | 0.567 | 21.32 | 0.537 | 22.39 | 0.672 | |
VDSR[ | 25.72 | 0.711 | 24.21 | 0.609 | 24.37 | 0.576 | 21.54 | 0.560 | 22.83 | 0.707 | |
SRGAN[ | 23.04 | 0.626 | 21.57 | 0.495 | 21.78 | 0.442 | 19.64 | 0.468 | 20.42 | 0.625 | |
LapSRN[ | 26.15 | 0.737 | 24.35 | 0.620 | 24.54 | 0.585 | 21.81 | 0.580 | 23.39 | 0.734 | |
ADSR[ | 25.60 | 0.710 | 24.18 | 0.600 | 24.31 | 0.572 | 21.40 | 0.552 | 22.75 | 0.698 | |
DBPN[ | 26.43 | 0.748 | 24.39 | 0.623 | 24.60 | 0.589 | 22.01 | 0.592 | 23.97 | 0.756 | |
IFANet (本文方法) | 26.90 | 0.770 | 24.70 | 0.635 | 24.70 | 0.590 | 22.25 | 0.605 | 24.69 | 0.779 |
表3 测试集上不同算法在不同放大因子下的PSNR及SSIM均值
Tab. 3 Mean values of PSNR/SSIM of different algorithms on test sets under different amplification factors
算法 | 放大 倍数 | Set5 | Set14 | BSD100 | Urban100 | Manga109 | |||||
---|---|---|---|---|---|---|---|---|---|---|---|
PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | PSNR/dB | SSIM | ||
Bicubic | 2× | 33.65 | 0.930 | 30.24 | 0.869 | 29.56 | 0.844 | 26.88 | 0.841 | 30.84 | 0.935 |
SRCNN[ | 36.66 | 0.954 | 32.45 | 0.906 | 31.36 | 0.888 | 29.52 | 0.895 | 35.72 | 0.968 | |
FSRCNN[ | 37.00 | 0.956 | 32.63 | 0.909 | 31.50 | 0.891 | 29.88 | 0.902 | — | — | |
VDSR[ | 37.74 | 0.959 | 32.97 | 0.913 | 31.90 | 0.896 | 30.77 | 0.914 | 37.16 | 0.974 | |
LapSRN[ | 37.52 | 0.929 | 33.08 | 0.913 | 31.80 | 0.895 | 31.05 | 0.910 | 37.53 | 0.974 | |
ADSR[ | 37.36 | 0.958 | 32.86 | 0.911 | 31.78 | 0.894 | 30.44 | 0.910 | — | — | |
DBPN[ | 37.60 | 0.959 | 33.18 | 0.914 | 31.94 | 0.897 | 31.20 | 0.918 | 37.57 | 0.974 | |
IFANet(本文方法) | 37.70 | 0.959 | 33.24 | 0.914 | 31.99 | 0.898 | 31.15 | 0.912 | 37.85 | 0.975 | |
Bicubic | 4× | 28.42 | 0.810 | 26.10 | 0.702 | 25.96 | 0.667 | 23.15 | 0.657 | 24.92 | 0.789 |
SRCNN[ | 30.49 | 0.862 | 27.61 | 0.751 | 26.91 | 0.710 | 24.53 | 0.722 | 27.66 | 0.858 | |
FSRCNN[ | 30.71 | 0.865 | 27.70 | 0.756 | 26.97 | 0.714 | 24.61 | 0.727 | 27.89 | 0.859 | |
VDSR[ | 31.53 | 0.883 | 28.03 | 0.767 | 27.29 | 0.725 | 25.18 | 0.752 | 28.82 | 0.886 | |
SRGAN[ | 29.46 | 0.838 | 26.60 | 0.718 | 25.74 | 0.666 | 24.50 | 0.736 | 27.79 | 0.856 | |
LapSRN[ | 31.54 | 0.885 | 28.09 | 0.770 | 27.31 | 0.727 | 25.21 | 0.756 | 29.09 | 0.890 | |
ADSR[ | 31.19 | 0.881 | 27.88 | 0.763 | 27.20 | 0.721 | 25.00 | 0.744 | — | — | |
DBPN[ | 31.76 | 0.887 | 28.39 | 0.778 | 27.48 | 0.733 | 25.71 | 0.772 | 30.22 | 0.902 | |
IFANet (本文方法) | 32.09 | 0.890 | 28.30 | 0.776 | 27.75 | 0.738 | 25.90 | 0.786 | 30.65 | 0.912 | |
Bicubic | 8× | 24.39 | 0.657 | 23.19 | 0.568 | 23.67 | 0.547 | 20.74 | 0.515 | 21.68 | 0.649 |
SRCNN[ | 25.33 | 0.689 | 23.85 | 0.593 | 24.13 | 0.565 | 21.29 | 0.543 | 22.37 | 0.682 | |
FSRCNN[ | 25.41 | 0.682 | 23.93 | 0.592 | 24.21 | 0.567 | 21.32 | 0.537 | 22.39 | 0.672 | |
VDSR[ | 25.72 | 0.711 | 24.21 | 0.609 | 24.37 | 0.576 | 21.54 | 0.560 | 22.83 | 0.707 | |
SRGAN[ | 23.04 | 0.626 | 21.57 | 0.495 | 21.78 | 0.442 | 19.64 | 0.468 | 20.42 | 0.625 | |
LapSRN[ | 26.15 | 0.737 | 24.35 | 0.620 | 24.54 | 0.585 | 21.81 | 0.580 | 23.39 | 0.734 | |
ADSR[ | 25.60 | 0.710 | 24.18 | 0.600 | 24.31 | 0.572 | 21.40 | 0.552 | 22.75 | 0.698 | |
DBPN[ | 26.43 | 0.748 | 24.39 | 0.623 | 24.60 | 0.589 | 22.01 | 0.592 | 23.97 | 0.756 | |
IFANet (本文方法) | 26.90 | 0.770 | 24.70 | 0.635 | 24.70 | 0.590 | 22.25 | 0.605 | 24.69 | 0.779 |
1 | DONG C, LOY C C, HE K M, et al. Image super-resolution using deep convolutional networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38(2): 295-307. 10.1109/tpami.2015.2439281 |
2 | SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[EB/OL]. (2015-04-10) [2022-06-01]. . |
3 | DONG C, LOY C C, TANG X O. Accelerating the super-resolution convolutional neural network [C]// Proceedings of the 2016 European Conference on Computer Vision, LNCS 9906. Cham: Springer, 2016: 391-407. |
4 | KIM J, LEE J K, LEE K M. Accurate image super-resolution using very deep convolutional networks [C]// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2016: 1646-1654. 10.1109/cvpr.2016.182 |
5 | LAI W S, HUANG J B, AHUJA N, et al. Deep Laplacian pyramid networks for fast and accurate super-resolution [C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2017: 5835-5843. 10.1109/cvpr.2017.618 |
6 | HARIS M, SHAKHNAROVICH G, UKITA N. Deep back-projection networks for super-resolution [C]// Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2018: 1664-1673. 10.1109/cvpr.2018.00179 |
7 | HU J, SHEN L, SUN G. Squeeze-and-excitation networks [C]// Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2018: 7132-7141. 10.1109/cvpr.2018.00745 |
8 | WANG F, JIANG M Q, QIAN C, et al. Residual attention network for image classification [C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2017: 6450-6458. 10.1109/cvpr.2017.683 |
9 | WANG X L, GIRSHICK R, GUPTA A, et al. Non-local neural networks [C]// Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2018: 7794-7803. 10.1109/cvpr.2018.00813 |
10 | LIU D, WEN B H, FAN Y C, et al. Non-local recurrent network for image restoration [C]// Proceedings of the 32nd International Conference on Neural Information Processing Systems. Red Hook, NY: Curran Associates Inc., 2018: 1680-1689. |
11 | ZHANG Y L, LI K P, LI K, et al. Image super-resolution using very deep residual channel attention networks [C]// Proceedings of the 2018 European Conference on Computer Vision, LNCS 11211. Cham: Springer, 2018: 294-310. |
12 | GOODFELLOW I J, POUGET-ABADIE J, MIRZA M, et al. Generative adversarial nets [C]// Proceedings of the 27th International Conference on Neural Information Processing Systems — Volume 2. Cambridge: MIT Press, 2014: 2672-2680. |
13 | JOHNSON J, ALAHI A, LI F F. Perceptual losses for real-time style transfer and super-resolution [C]// Proceedings of the 2016 European Conference on Computer Vision, LNCS 9906. Cham: Springer, 2016: 694-711. |
14 | LEDIG C, THEIS L, HUSZÁR F, et al. Photo-realistic single image super-resolution using a generative adversarial network [C]// Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2017: 105-114. 10.1109/cvpr.2017.19 |
15 | SHI W Z, CABALLERO J, HUSZÁR F, et al. Real-time single image and video super-resolution using an efficient sub-pixel convolutional neural network [C]// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2016: 1874-1883. 10.1109/cvpr.2016.207 |
16 | HE K M, ZHANG X Y, REN S Q, et al. Deep residual learning for image recognition [C]// Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2016: 770-778. 10.1109/cvpr.2016.90 |
17 | CHANG J R, CHEN Y S. Pyramid stereo matching network [C]// Proceedings of the 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2018: 5410-5418. 10.1109/cvpr.2018.00567 |
18 | ANWAR S, KHAN S, BARNES N. A deep journey into super-resolution: a survey[J]. ACM Computing Surveys, 2020, 53(3): No.60. 10.1145/3390462 |
19 | ANWAR S, BARNES N. Densely residual Laplacian super-resolution[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 44(3): 1192-1204. 10.1109/tpami.2020.3021088 |
20 | BELL S, UPCHURCH P, SNAVELY N, et al. Material recognition in the wild with the materials in context database [C]// Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2015: 3479-3487. 10.1109/cvpr.2015.7298970 |
21 | 梁敏,王昊榕,张瑶,等.基于加速残差网络的图像超分辨率重建方法[J].计算机应用, 2021, 41(5): 1438-1444. |
LIANG M, WANG H R, ZHANG Y, et al. Image super-resolution reconstruction method based on accelerated residual network[J]. Journal of Computer Applications, 2021, 41(5): 1438-1444. |
[1] | 潘烨新, 杨哲. 基于多级特征双向融合的小目标检测优化模型[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2871-2877. |
[2] | 赵志强, 马培红, 黑新宏. 基于双重注意力机制的人群计数方法[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2886-2892. |
[3] | 李顺勇, 李师毅, 胥瑞, 赵兴旺. 基于自注意力融合的不完整多视图聚类算法[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2696-2703. |
[4] | 黄云川, 江永全, 黄骏涛, 杨燕. 基于元图同构网络的分子毒性预测[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2964-2969. |
[5] | 秦璟, 秦志光, 李发礼, 彭悦恒. 基于概率稀疏自注意力神经网络的重性抑郁疾患诊断[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2970-2974. |
[6] | 王熙源, 张战成, 徐少康, 张宝成, 罗晓清, 胡伏原. 面向手术导航3D/2D配准的无监督跨域迁移网络[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2911-2918. |
[7] | 李力铤, 华蓓, 贺若舟, 徐况. 基于解耦注意力机制的多变量时序预测模型[J]. 《计算机应用》唯一官方网站, 2024, 44(9): 2732-2738. |
[8] | 薛凯鹏, 徐涛, 廖春节. 融合自监督和多层交叉注意力的多模态情感分析网络[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2387-2392. |
[9] | 汪雨晴, 朱广丽, 段文杰, 李书羽, 周若彤. 基于交互注意力机制的心理咨询文本情感分类模型[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2393-2399. |
[10] | 高鹏淇, 黄鹤鸣, 樊永红. 融合坐标与多头注意力机制的交互语音情感识别[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2400-2406. |
[11] | 刘禹含, 吉根林, 张红苹. 基于骨架图与混合注意力的视频行人异常检测方法[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2551-2557. |
[12] | 李钟华, 白云起, 王雪津, 黄雷雷, 林初俊, 廖诗宇. 基于图像增强的低照度人脸检测[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2588-2594. |
[13] | 莫尚斌, 王文君, 董凌, 高盛祥, 余正涛. 基于多路信息聚合协同解码的单通道语音增强[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2611-2617. |
[14] | 顾焰杰, 张英俊, 刘晓倩, 周围, 孙威. 基于时空多图融合的交通流量预测[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2618-2625. |
[15] | 石乾宏, 杨燕, 江永全, 欧阳小草, 范武波, 陈强, 姜涛, 李媛. 面向空气质量预测的多粒度突变拟合网络[J]. 《计算机应用》唯一官方网站, 2024, 44(8): 2643-2650. |
阅读次数 | ||||||
全文 |
|
|||||
摘要 |
|
|||||