帮助 关于我们

返回检索结果

基于SPL迭代思想的图像压缩感知重构神经网络
Image Compressive Sensing Reconstruction Network Based on Iterative SPL Theory

查看参考文献24篇

裴翰奇 1   杨春玲 1 *   魏志超 1   曹燕 2  
文摘 由于神经网络强大的学习能力与快速的运行速度,近年来基于深度学习的图像压缩感知(Image Compressive Sensing, ICS)研究备受关注.然而,大多数现有ICS神经网络的结构设计忽略了传统迭代重构算法中的数学理论基础,无法有效利用信号中的先验结构知识,可解释性较差.为了保留优化算法核心思想并同时利用深度学习的高性能,本文使用可学习的卷积层替代了传统平滑投影Landweber算法(Smooth Projected Landweber,SPL)中的人工设计参数,提出一种新型ICS神经网络SPLNet.在SPLNet中,设计了一个独特的网络结构SPLBlock实现SPL迭代过程中的三个核心步骤:(1)去除块效应的维纳滤波器;(2)在凸投影集合上的近似操作;(3)实现稀疏表示及去噪的变换域双变量收缩.仿真实验结果表明:与现有最优的ICS优化迭代算法GSR相比,SPLNet的重构图像平均PSNR提升了0.78dB;与最优的神经网络框架SCSNet相比,SPLNet的重构图像平均PSNR提升了0.92dB.
其他语种文摘 Due to its great learning ability and fast processing speed,deep learning-based image compressive sensing (ICS) methods attract a lot of attention in recent years. However, the design of most existing ICS neural networks architecture ignore the mathematical theory in iterative optimization-based methods and cannot effectively use the prior structure knowledge in the signal, leading to lack of the interpretability. In order to retain the core ideas of the optimization algorithm and utilize the high performance of deep learning, this paper uses learnable convolutional layers to replace the predefined filters and artificial design parameters in the traditional smooth projected Landweber algorithm (SPL),and proposes a ICS neural network named SPLNet. In SPLNet,we design a unique network structure SPLBlock to implement three key steps in SPL iteration: (1) Wiener filter for removal of blocking artifacts; (2) approximation with projection onto the convex set; (3) bivariate shrinkage on transform domain for sparse representation and denoising. Experimental results indicate that, compared with current state-of-the-art ICS optimization iterative algorithm GSR, the average reconstructed image PSNR of SPLNet are improved by 0.78dB, and compared with state-of-the-art neural network framework SCSNet, the average reconstructed image PSNR of SPLNet are improved by 0.92dB.
来源 电子学报 ,2021,49(6):1195-1203 【核心库】
DOI 10.12263/DZXB.20200618
关键词 压缩感知 ; 图像压缩感知 ; 深度网络 ; 卷积神经网络 ; 图像采样 ; 图像重构
地址

1. 华南理工大学电子与信息学院, 广东, 广州, 510640  

2. 华南理工大学, 国家移动超声探测工程技术研究中心, 广东, 广州, 510640

语种 中文
文献类型 研究性论文
ISSN 0372-2112
学科 电子技术、通信技术
基金 广东省自然科学基金重点项目 ;  广东省“短距离无线探测与通信”重点实验室 ;  国家移动超声探测工程技术研究中心
文献收藏号 CSCD:7018852

参考文献 共 24 共2页

1.  Donoho D L. Compressed sensing. IEEE Transactions on Information Theory,2006,52(4):1289-1306 CSCD被引 2930 次    
2.  Gan L. Block compressed sensing of natural images. 2007 15th International Conference on Digital Signal Processing,2007:403-406 CSCD被引 10 次    
3.  Mun S. Block compressed sensing of images using directional transforms. 2009 16th IEEE International Conference on Image Processing (ICIP),2009:3021-3024 CSCD被引 2 次    
4.  Fowler J E. Multiscale block compressed sensing with smoothed projected landweber reconstruction. 2011 19th European Signal Processing Conference,2011:564-568 CSCD被引 2 次    
5.  Mun S. Residual reconstruction for block-based compressed sensing of video. 2011 Data Compression Conference,2011:183-192 CSCD被引 4 次    
6.  Chen C. Compressed-sensing recovery of images and video using multihypothesis predictions. 2011 Conference Record of the Forty Fifth Asilomar Conference on Signals,Systems and Computers (ASILOMAR),2011:1193-1198 CSCD被引 2 次    
7.  Zhang J. Group-based sparse representation for image restoration. IEEE Transactions on Image Processing,2014,23(8):3336-3351 CSCD被引 84 次    
8.  Fan Q. Few-shot object detection with attention-RPN and multi-relation detector. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,2020:4013-4022 CSCD被引 8 次    
9.  Chen Z. Siamesebox adaptive network for visual tracking. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,2020:6668-6677 CSCD被引 9 次    
10.  Xiang X. Zooming Slow-Mo: Fast and accurate one-stage space-time video super-resolution. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition,2020:3370-3379 CSCD被引 2 次    
11.  Kulkarni K. ReconNet: Non-iterative reconstruction of images from compressively sensed measurements. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2016:449-458 CSCD被引 9 次    
12.  Yao H. DR2-Net: Deep residual reconstruction network for image compressive sensing. Neurocomputing,2019,359:483-493 CSCD被引 13 次    
13.  Zhang J. ISTA-Net: Interpretable optimizationinspired deep network for image compressive sensing. Proceedings of the IEEEConference on Computer Vision and Pattern Recognition,2018:1828-1837 CSCD被引 1 次    
14.  练秋生. 基于多尺度残差网络的压缩感知重构算法. 自动化学报,2019,45(11):2082-2091 CSCD被引 13 次    
15.  Shi W. Deep networks for compressed image sensing. 2017 IEEE International Conference on Multimedia and Expo (ICME),2017:877-882 CSCD被引 2 次    
16.  Shi W. Multi-scale deep networks for image compressed sensing. 2018 25th IEEE International Conference on Image Processing (ICIP),2018:46-50 CSCD被引 1 次    
17.  Shi W. Image Compressed sensing using convolutional neural network. IEEE Transactions on Image Processing,2019,29:375-388 CSCD被引 4 次    
18.  Shi W. Scalable convolutional neural network for image compressed sensing. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition,2019:12290-12299 CSCD被引 1 次    
19.  Cui W. Deep neural network based sparse measurement matrix for image compressed sensing. 2018 25th IEEE International Conference on Image Processing (ICIP),2018:3883-3887 CSCD被引 2 次    
20.  He K. Delving deep into rectifiers: Surpassing human-level performance onimagenet classification. Proceedings of the IEEE International Conference on Computer Vision,2015:1026-1034 CSCD被引 112 次    
引证文献 3 篇

1 曾春艳 面向可解释压缩感知的算法展开综述 华中科技大学学报. 自然科学版,2022,50(11):35-43
CSCD被引 0 次

2 魏志超 时域注意力特征对齐的视频压缩感知重构网络 电子学报,2022,50(11):2584-2592
CSCD被引 2 次

显示所有3篇文献

论文科学数据集
PlumX Metrics
相关文献

 作者相关
 关键词相关
 参考文献相关

版权所有 ©2008 中国科学院文献情报中心 制作维护：中国科学院文献情报中心
地址：北京中关村北四环西路33号 邮政编码：100190 联系电话：(010)82627496 E-mail:cscd@mail.las.ac.cn 京ICP备05002861号-4 | 京公网安备11010802043238号