一级黄色片免费播放|中国黄色视频播放片|日本三级a|可以直接考播黄片影视免费一级毛片

高級搜索

留言板

尊敬的讀者、作者、審稿人, 關(guān)于本刊的投稿、審稿、編輯和出版的任何問題, 您可以本頁添加留言。我們將盡快給您答復(fù)。謝謝您的支持!

姓名
郵箱
手機號碼
標(biāo)題
留言內(nèi)容
驗證碼

基于迭代交替優(yōu)化的圖像盲超分辨率重建

陳洪剛 李自強 張永飛 王正勇 卿粼波 何小海

陳洪剛, 李自強, 張永飛, 王正勇, 卿粼波, 何小海. 基于迭代交替優(yōu)化的圖像盲超分辨率重建[J]. 電子與信息學(xué)報, 2022, 44(10): 3343-3352. doi: 10.11999/JEIT220380
引用本文: 陳洪剛, 李自強, 張永飛, 王正勇, 卿粼波, 何小海. 基于迭代交替優(yōu)化的圖像盲超分辨率重建[J]. 電子與信息學(xué)報, 2022, 44(10): 3343-3352. doi: 10.11999/JEIT220380
CHEN Honggang, LI Ziqiang, ZHANG Yongfei, WANG Zhengyong, QING Linbo, HE Xiaohai. Blind Image Super-resolution Reconstruction via Iterative and Alternative Optimization[J]. Journal of Electronics & Information Technology, 2022, 44(10): 3343-3352. doi: 10.11999/JEIT220380
Citation: CHEN Honggang, LI Ziqiang, ZHANG Yongfei, WANG Zhengyong, QING Linbo, HE Xiaohai. Blind Image Super-resolution Reconstruction via Iterative and Alternative Optimization[J]. Journal of Electronics & Information Technology, 2022, 44(10): 3343-3352. doi: 10.11999/JEIT220380

基于迭代交替優(yōu)化的圖像盲超分辨率重建

doi: 10.11999/JEIT220380
基金項目: 國家自然科學(xué)基金(62001316, 61871279),四川省自然科學(xué)基金(2022NSFSC0922),中央高?;究蒲袠I(yè)務(wù)費專項資金(2021SCU12061)
詳細(xì)信息
    作者簡介:

    陳洪剛:男,副研究員,研究方向為圖像與視頻處理

    李自強:男,碩士生,研究方向為圖像超分辨率重建

    張永飛:男,碩士生,研究方向為圖像超分辨率重建

    王正勇:女,副教授,研究方向為圖像與視頻處理

    卿粼波:男, 教授,研究方向為圖像與視頻處理

    何小海:男,教授,研究方向為圖像與視頻處理

    通訊作者:

    王正勇 wangzheny@scu.edu.cn

  • 中圖分類號: TN911.73; TP391

Blind Image Super-resolution Reconstruction via Iterative and Alternative Optimization

Funds: The National Natural Science Foundation of China (62001316, 61871279), The Natural Science Foundation of Sichuan Province (2022NSFSC0922), The Fundamental Research Foundation for the Central Universities (2021SCU12061)
  • 摘要: 基于深度卷積神經(jīng)網(wǎng)絡(luò)的圖像超分辨率重建算法通常假設(shè)低分辨率圖像的降質(zhì)是固定且已知的,如雙3次下采樣等,因此難以處理降質(zhì)(如模糊核及噪聲水平)未知的圖像。針對此問題,該文提出聯(lián)合估計模糊核、噪聲水平和高分辨率圖像,設(shè)計了一種基于迭代交替優(yōu)化的圖像盲超分辨率重建網(wǎng)絡(luò)。在所提網(wǎng)絡(luò)中,圖像重建器以估計的模糊核和噪聲水平作為先驗信息,由低分辨率圖像重建出高分辨率圖像;同時,綜合低分辨率圖像和估計的高分辨率圖像,模糊核及噪聲水平估計器分別實現(xiàn)模糊核和噪聲水平的估計。進一步地,該文提出對模糊核/噪聲水平估計器及圖像重建器進行迭代交替的端對端優(yōu)化,以提高它們的兼容性并使其相互促進。實驗結(jié)果表明,與IKC, DASR, MANet, DAN等現(xiàn)有算法相比,提出方法在常用公開測試集(Set5, Set14, B100, Urban100)及真實場景圖像上都取得了更優(yōu)的性能,能夠更好地對降質(zhì)未知的圖像進行重建;同時,提出方法在參數(shù)量或處理效率上也有一定的優(yōu)勢。
  • 圖  1  提出算法的原理框圖

    圖  2  構(gòu)建的圖像重建器、模糊核估計器及噪聲水平估計器的網(wǎng)絡(luò)結(jié)構(gòu)

    圖  3  動態(tài)調(diào)制層(DML)

    圖  4  動態(tài)注意力模塊(DAB)

    圖  5  不同算法對Urban100中“img097”圖像的重建結(jié)果

    圖  6  不同算法對真實場景圖像“chip”的重建結(jié)果(重建尺度為4)

    圖  7  Set5中圖像的重建結(jié)果、模糊核估計及噪聲水平估計隨迭代次數(shù)的動態(tài)變化過程

    圖  8  不同迭代次數(shù)下對Set5中“baby”圖像的重建結(jié)果

    圖  9  參數(shù)量與運行時間

    表  1  2倍重建結(jié)果的客觀參數(shù)PSNR(dB)/SSIM比較

    方法噪聲水平Set5[27]Set14[28]B100[29]Urban100[30]
    Bicubic530.07/0.844227.61/0.762027.23/0.726224.61/0.7253
    1028.85/0.770926.85/0.697926.51/0.660124.19/0.6637
    MANet[12]533.60/0.910130.53/0.840729.45/0.805628.31/0.8513
    1032.15/0.887129.42/0.804828.40/0.762927.31/0.8202
    DASR[14]533.35/0.907830.22/0.832529.12/0.794727.66/0.8364
    1031.95/0.885529.21/0.800328.21/0.756326.89/0.8107
    DnCNN[31]+IKC[15]531.69/0.882429.27/0.811828.67/0.782626.79/0.8089
    1030.85/0.865228.53/0.781727.90/0.746326.12/0.7830
    DAN[17]533.83/0.913230.69/0.843029.52/0.808128.62/0.8566
    1032.32/0.890229.53/0.807228.45/0.765027.56/0.8256
    本文算法533.98/0.915330.85/0.849229.66/0.814028.79/0.8609
    1032.39/0.891229.64/0.811028.54/0.768327.68/0.8283
    下載: 導(dǎo)出CSV

    表  2  4倍重建結(jié)果的客觀參數(shù)PSNR(dB)/SSIM比較

    方法噪聲水平Set5[27]Set14[28]B100[29]Urban100[30]
    Bicubic525.84/0.716224.29/0.614424.53/0.582521.70/0.5644
    1025.30/0.672823.91/0.576824.12/0.543821.48/0.5286
    MANet[12]529.01/0.824226.59/0.698326.01/0.650724.01/0.6922
    1027.77/0.792825.74/0.667225.37/0.620723.36/0.6605
    DASR[14]528.85/0.821426.46/0.696625.94/0.649423.72/0.6880
    1027.73/0.792125.69/0.667625.32/0.620723.16/0.6605
    DnCNN[31]+IKC[15]527.26/0.761925.51/0.660425.38/0.621822.92/0.6332
    1026.65/0.749325.03/0.639024.96/0.600822.53/0.6140
    DAN[17]529.01/0.823826.62/0.698026.02/0.650724.02/0.6903
    1027.84/0.794725.85/0.669825.39/0.622323.44/0.6630
    本文算法529.32/0.830026.82/0.706026.16/0.658424.29/0.7036
    1028.05/0.799925.98/0.673825.48/0.626023.61/0.6712
    下載: 導(dǎo)出CSV
  • [1] 蔡文郁, 張美燕, 吳巖, 等. 基于循環(huán)生成對抗網(wǎng)絡(luò)的超分辨率重建算法研究[J]. 電子與信息學(xué)報, 2022, 44(1): 178–186. doi: 10.11999/JEIT201046

    CAI Wenyu, ZHANG Meiyan, WU Yan, et al. Research on cyclic generation countermeasure network based super-resolution image reconstruction algorithm[J]. Journal of Electronics &Information Technology, 2022, 44(1): 178–186. doi: 10.11999/JEIT201046
    [2] ZHANG Xiangjun and WU Xiaolin. Image interpolation by adaptive 2-D autoregressive modeling and soft-decision estimation[J]. IEEE Transactions on Image Processing, 2008, 17(6): 887–896. doi: 10.1109/TIP.2008.924279
    [3] ZHANG Kaibing, GAO Xinbo, TAO Dacheng, et al. Single image super-resolution with non-local means and steering kernel regression[J]. IEEE Transactions on Image Processing, 2012, 21(11): 4544–4556. doi: 10.1109/TIP.2012.2208977
    [4] DONG Chao, LOY C C, HE Kaiming, et al. Learning a deep convolutional network for image super-resolution[C]. Proceedings of the 13th European Conference on Computer Vision, Zurich, Switzerland, 2014: 184–199.
    [5] CHEN Hanting, WANG Yunhe, GUO Tianyu, et al. Pre-trained image processing transformer[C]. Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, USA, 2021: 12299–12310.
    [6] LIANG Jingyun, CAO Jiezhang, SUN Guolei, et al. SwinIR: Image restoration using swin transformer[C]. Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision Workshops, Montreal, Canada, 2021: 1833–1844.
    [7] ZHANG Kai, LIANG Jingyun, VAN GOOL L, et al. Designing a practical degradation model for deep blind image super-resolution[C]. Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision, Montreal, Canada, 2021: 4791–4800.
    [8] WANG Xintao, XIE Liangbin, DONG Chao, et al. Real-ESRGAN: Training real-world blind super-resolution with pure synthetic data[C]. Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision, Montreal, Canada, 2021: 1905–1914.
    [9] BELL-KLIGLER S, SHOCHER A, and IRANI M. Blind super-resolution kernel estimation using an internal-GAN[C]. Proceedings of the 33rd International Conference on Neural Information Processing Systems, Vancouver, Canada, 2019: 284–293.
    [10] LIANG Jingyun, ZHANG Kai, GU Shuhang, et al. Flow-based kernel prior with application to blind super-resolution[C]. Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, USA, 2021: 10601–10610.
    [11] TAO Guangpin, JI Xiaozhong, WANG Wenzhuo, et al. Spectrum-to-kernel translation for accurate blind image super-resolution[C/OL]. Advances in Neural Information Processing Systems, 2021: 34.
    [12] LIANG Jingyun, SUN Guolei, ZHANG Kai, et al. Mutual affine network for spatially variant kernel estimation in blind image super-resolution[C]. Proceedings of the 2021 IEEE/CVF International Conference on Computer Vision, Montreal, Canada, 2021: 4096–4105.
    [13] KIM J, JUNG C, and KIM C. Dual back-projection-based internal learning for blind super-resolution[J]. IEEE Signal Processing Letters, 2020, 27: 1190–1194. doi: 10.1109/LSP.2020.3005043
    [14] WANG Longguang, WANG Yingqian, DONG Xiaoyu, et al. Unsupervised degradation representation learning for blind super-resolution[C]. Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, USA, 2021: 10581–10590.
    [15] GU Jinjin, LU Hannan, ZUO Wangmeng, et al. Blind super-resolution with iterative kernel correction[C]. Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, USA, 2019: 1604–1613.
    [16] LUO Zhengxiong, HUANG Yan, LI Shang, et al. Unfolding the alternating optimization for blind super resolution[C/OL]. Advances in Neural Information Processing Systems, 2020: 5632–5643.
    [17] LUO Zhengxiong, HUANG Yan, LI Shang, et al. End-to-end alternating optimization for blind super resolution[EB/OL]. https://arxiv.org/abs/2105.06878v1, 2021.
    [18] WANG Zhihao, CHEN Jian, and HOI S C H. Deep learning for image super-resolution: A survey[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 43(10): 3365–3387. doi: 10.1109/TPAMI.2020.2982166
    [19] CHEN Honggang, HE Xiaohai, QING Linbo, et al. Real-world single image super-resolution: A brief review[J]. Information Fusion, 2022, 79: 124–145. doi: 10.1016/j.inffus.2021.09.005
    [20] LIU Anran, LIU Yihao, GU Jinjin, et al. Blind image super-resolution: A survey and beyond[EB/OL]. https://arxiv.org/abs/2107.03055, 2021.
    [21] HUI Zheng, LI Jie, WANG Xiumei, et al. Learning the non-differentiable optimization for blind super-resolution[C]. Proceedings of the 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, USA, 2021: 2093–2102.
    [22] CHEN Haoyu, GU Jinjin, and ZHANG Zhi. Attention in attention network for image super-resolution[EB/OL]. https://arxiv.org/abs/2104.09497v3, 2021.
    [23] GUO Shi, YAN Zifei, ZHANG Kai, et al. Toward convolutional blind denoising of real photographs[C]. Proceedings of the 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, USA, 2019: 1712–1722.
    [24] AGUSTSSON E and TIMOFTE R. NTIRE 2017 challenge on single image super-resolution: Dataset and study[C]. Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops, Honolulu, USA, 2017: 126–135.
    [25] TIMOFTE R, AGUSTSSON E, VAN GOOL L, et al. NTIRE 2017 challenge on single image super-resolution: Methods and results[C]. Proceedings of the 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops, Honolulu, USA, 2017: 114–125.
    [26] KINGMA D P and BA J. ADAM: A method for stochastic optimization[C]. 3rd International Conference on Learning Representations, San Diego, USA, 2014.
    [27] BEVILACQUA M, ROUMY A, GUILLEMOT C, et al. Low-complexity single-image super-resolution based on nonnegative neighbor embedding[C]. British Machine Vision Conference, Surrey, United Kingdom, 2012.
    [28] ZEYDE R, ELAD M, and PROTTER M. On single image scale-up using sparse-representations[C]. 7th International Conference on Curves and Surfaces, Avignon, France, 2010: 711–730.
    [29] MARTIN D, FOWLKES C, TAL D, et al. A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics[C]. Proceedings Eighth IEEE International Conference on Computer Vision, Vancouver, Canada, 2001: 416–423.
    [30] HUANG J B, SINGH A, and AHUJA N. Single image super-resolution from transformed self-exemplars[C]. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, Boston, USA, 2015: 5197–5206.
    [31] ZHANG Kai, ZUO Wangmeng, CHEN Yunjin, et al. Beyond a gaussian denoiser: Residual learning of deep CNN for image denoising[J]. IEEE Transactions on Image Processing, 2017, 26(7): 3142–3155. doi: 10.1109/TIP.2017.2662206
  • 加載中
圖(9) / 表(2)
計量
  • 文章訪問數(shù):  1098
  • HTML全文瀏覽量:  642
  • PDF下載量:  273
  • 被引次數(shù): 0
出版歷程
  • 收稿日期:  2022-04-01
  • 修回日期:  2022-05-21
  • 網(wǎng)絡(luò)出版日期:  2022-07-01
  • 刊出日期:  2022-10-19

目錄

    /

    返回文章
    返回