BDCN和U-net边缘生成两阶段修复算法

doi:10.13190/j.jbupt.2020-268

北京邮电大学学报 ›› 2021, Vol. 44 ›› Issue (5): 121-126.doi: 10.13190/j.jbupt.2020-268

BDCN和U-net边缘生成两阶段修复算法

李海燕¹, 王伟华¹, 郭磊¹, 李海江², 李红松¹

1. 云南大学信息学院, 昆明 650500;
2. 云南交通投资建设集团有限公司, 昆明 650299

收稿日期:2020-07-01 出版日期:2021-10-28 发布日期:2021-09-06
通讯作者: 郭磊(1974-),男,高级实验师,E-mail:guolei@ynu.edu.cn. E-mail:guolei@ynu.edu.cn
作者简介:李海燕(1976-),女,教授,博士生导师.
基金资助:
国家自然科学基金项目（61861045）；云南省基础研究计划重点项目（202101AS0070031）

Two-Stage Network Inpainting Algorithm Based on BDCN and U-net Edge Generation

LI Hai-yan¹, WANG Wei-hua¹, GUO Lei¹, LI Hai-jiang², LI Hong-song¹

1. School of Information Science and Engineering, Yunnan University, Kunming 650500, China;
2. Yunnan Communications Investment and Construction Group Company Limited, Kunming 650299, China

Received:2020-07-01 Online:2021-10-28 Published:2021-09-06

摘要/Abstract

摘要： 为了对图像中大面积不规则缺失区域进行合理的结构修复和精细的纹理重构，提出了一种基于双向级联边缘检测网络（BDCN）和U-net残缺边缘生成的两阶段网络图像修复算法.第一阶段首先基于BDCN网络提取边缘，然后基于U-net架构的边缘生成网络用下采样对缺失图像边缘提取特征，再结合上采样的输入信息和下采样各层信息还原图像边缘纹理细节；第二阶段使用空洞卷积进行下采样和上采样，经过残差网络重建细节丰富的缺失图像.在公开数据集上将所提算法与现有经典算法进行对比，实验结果表明，所提算法能得到合理的结构和精细的纹理细节，优于对比算法.

关键词: 图像修复, 双向级联边缘检测网络边缘提取, U-net残缺边缘生成器, 两阶段网络

Abstract: To repair the large irregular missing areas of an image, and obtain reasonable structure and fine-detailed textures,a two-stage network image inpainting algorithm based on bi-directional cascade network (BDCN) for perceptual edge detection and U-net incomplete edge generation is proposed. Firstly,the algorithm edges are extracted using the BDCN network. In the first stage,down-sampling is used to extract the features of the missing image edges based on the edge generation network of the U-net network architecture. The information inputted by the up-sampling layer and the information of each down-sampling layer is then combined to restore the image edge texture details. In the second stage,the hole convolution is applied for down-sampling and up-sampling,which adopts the residual network to reconstruct the missing image with rich details. The proposed algorithm is compared with the existing classic algorithm on the public datasets. Experimental results demonstrate that the proposed algorithm can obtain reasonable results and fine texture details,and its performance is superior to those of the contrast algorithms.

Key words: image inpainting, bi-directional cascade network for perceptual edge detection, U-net incomplete edge generator, the two-stage network

中图分类号:

TP391

李海燕, 王伟华, 郭磊, 李海江, 李红松. BDCN和U-net边缘生成两阶段修复算法[J]. 北京邮电大学学报, 2021, 44(5): 121-126.

LI Hai-yan, WANG Wei-hua, GUO Lei, LI Hai-jiang, LI Hong-song. Two-Stage Network Inpainting Algorithm Based on BDCN and U-net Edge Generation[J]. Journal of Beijing University of Posts and Telecommunications, 2021, 44(5): 121-126.

参考文献

[1] Qiu Fudong, Wu Fan, Chen Guihai. Privacy and quality preserving multimedia data aggregation for participatory sensing systems[J]. IEEE Transactions on Mobile Computing, 2014, 14(6):1287-1300.
[2] Rahaman S, Cheng L, Yao D D, et al. Provably secure anonymous-yet-accountable crowdsensing with scalable sublinear revocation[J]. Privacy Enhancing Technologies, 2017(4):384-403.
[3] Sucasas V, Mantas G, Bastos J, et al. A signature scheme with un-linkable-yet-accountable pseudonymity for privacy-preserving crowdsensing[J]. IEEE Transactions on Mobile Computing, 2019, 19(4):752-768.
[4] Ni Jianbing, Zhang Kuan, Xia Qi, et al. Enabling strong privacy preservation and accurate task allocation for mobile crowdsensing[J]. IEEE Transactions on Mobile Computing, 2019, 19(6):1317-1331.
[5] Zhao Bowen, Tang Shaohua, Liu Ximeng, et al. PACE:privacy-preserving and quality-aware incentive mechanism for mobile crowdsensing[J]. IEEE Transactions on Mobile Computing, 2020, 20(5):1924-1939.
[6] Sun Peng, Wang Zhibo, Feng Yunhe, et al. Towards personalized privacy-preserving incentive for truth disco-very in crowdsourced binary-choice question answering[C]//IEEE Conference on Computer Communications. Toronto:IEEE, 2020:1133-1142.
[7] Galbraith S D, Paterson K G, Smart N P. Pairings for cryptographers[J].Discrete Applied Mathematics, 2008, 156(16):3113-3121.
[8] Chase M, Lysyanskaya A. On signatures of knowledge[C]//Annual International Cryptology Conference. Berlin:Springer, 2006:78-96.
[9] Camenisch J. Group signature schemes and payment systems based on the discrete logarithm problem[D]. Zurich:Swiss Federal Institute of Technology, 1998.
[10] Pointcheval D, Sanders O. Short randomizable signatures[C]//Cryptographers' Track at the RSA Conference. San Francisco:Springer, 2016:111-126.
[11] Camenisch J, Lysyanskaya A. Dynamic accumulators and application to efficient revocation of anonymous credentials[C]//Annual International Cryptology Conference. Berlin:Springer, 2002:61-76.
[12] Baldimtsi F, Camenisch J, Dubovitskaya M, et al. Accumulators with applications to anonymity-preserving revocation[C]//IEEE European Symposium on Security and Privacy. Paris:IEEE, 2017:301-315.
[13] Camenisch J, Drijvers M, Lehmann A, et al. Short threshold dynamic group signatures[C]//International Conference on Security and Cryptography for Networks. Amalfi:Springer, 2020:401-423.
[14] MIRACL Company Limited. Multiprecision integer and rational arithmetic cryptographic library version 7.0[EB/OL]. (2019-08-21)[2020-10-20]. https://github.com/miracl/MIRACL.
[15] Dang Q H. Secure Hash standard:NIST FIPS 180-4-2015[S]. Gaithersburg:National Institute of Standards and Technology, 2015:21-23.
[16] Fan J, Vercauteren F, Verbauwhede I. Faster F_p-arithmetic for cryptographic pairings on Barreto-Naehrig curves[C]//International Workshop on Cryptographic Hardware and Embedded Systems. Berlin:Springer, 2009:240-253.

BDCN和U-net边缘生成两阶段修复算法

Two-Stage Network Inpainting Algorithm Based on BDCN and U-net Edge Generation

RichHTML

可视化

摘要/Abstract

引用本文

使用本文

参考文献

相关文章 2

编辑推荐

Metrics

本文评价

[1]	李海燕晁艳静余鹏飞李海江张榆锋. 上下文信息多样聚合的图像修复算法[J]. 北京邮电大学学报, 2023, 46(1): 19-25.
[2]	曹志义, 牛少彰, 张继威. 基于生成对抗网络的遮挡图像修复算法[J]. 北京邮电大学学报, 2018, 41(3): 81-86.