基于深度学习SDA的压缩感知图像重构方法
|
摘要
为减少传统压缩感知对图像信号测量与重构的时间,提高重构精度,提出一种深度学习SDA的压缩感知框架,采用堆叠去噪自动编码器(SDA)作为无监督特征学习器,支持信号的线性和非线性测量,捕获特定信号的不同元素之间的统计依赖性;利用前馈深度神经网络代替传统重构算法,从训练数据中学习信号的结构化表示,实现图像信号重构。实验结果表明,与压缩感知SPL算法、D-AMP算法和TV算法比较,该方法峰值信噪比(PSNR)更高,且重构时间仅仅为0.002s,远低于其它3种算法。
To decrease the measurement and reconstruction time and improve the reconstruction accuracy of traditional compressive sensing,a compressed sensing framework based on deep learning SDA was proposed.Stacked denoising autoencoder(SDA)was used as an unsupervised feature learner,in which linear and non-linear measurements of the signal were supported,to capture the statistical dependencies between different elements of a given signal.The feed-forward neural network was used to replace the traditional reconstruction algorithm to learn the structured representation of the signal from the training data and to reconstruct the image signal.Experimental results show that the PSNR of the proposed method is higher than that of compressed sensing SPL algorithm,D-AMP algorithm and TV algorithm.And the reconstruction time is only 0.002 seconds,which is much lower than the other three algorithms.
To decrease the measurement and reconstruction time and improve the reconstruction accuracy of traditional compressive sensing,a compressed sensing framework based on deep learning SDA was proposed.Stacked denoising autoencoder(SDA)was used as an unsupervised feature learner,in which linear and non-linear measurements of the signal were supported,to capture the statistical dependencies between different elements of a given signal.The feed-forward neural network was used to replace the traditional reconstruction algorithm to learn the structured representation of the signal from the training data and to reconstruct the image signal.Experimental results show that the PSNR of the proposed method is higher than that of compressed sensing SPL algorithm,D-AMP algorithm and TV algorithm.And the reconstruction time is only 0.002 seconds,which is much lower than the other three algorithms.
引文
[1]Zhang Xuepan,Liao Guisheng,Zhu Shengqi,et al.Efficient compressed sensing method for moving-target imaging by exploiting the geometry information of the defocused results[J].IEEE Geoscience and Remote Sensing Letters,2015,12(3):517-521.
[2]Metzler C A,Maleki A,Baraniuk R G.From denoising to compressed sensing[J].IEEE Transactions on Information Theory,2016,62(9):5117-5144.
[3]LI Nan,REN Qinghua,SU Yuze.Constructive method of CS measurement matrix in TDCS[J].Computer Engineering and Design,2017,38(1):7-11(in Chinese).[李楠,任清华,苏玉泽.TDCS压缩感知观测矩阵构造方法[J].计算机工程与设计,2017,38(1):7-11.]
[4]Voroninski V,Xu Z.A strong restricted isometry property,with an application to phaseless compressed sensing[J].Applied and Computational Harmonic Analysis,2016,40(2):386-395.
[5]Tillmann A M,Pfetsch M E.The computational complexity of the restricted isometry property,the nullspace property,and related concepts in compressed sensing[J].IEEE Transactions on Information Theory,2014,60(2):1248-1259.
[6]Zhou Nanrun,Zhang Aidi,Zheng Fen,et al.Novel image compression-encryption hybrid algorithm based on key-controlled measurement matrix in compressive sensing[J].Optics&Laser Technology,2014,62(10):152-160.
[7]MA Riliang,PEI Liye,JIANG Hua.Improved optimization algorithm for measurement matrix in compressed sensing[J].Journal of Signal Processing,2017,33(2):192-197(in Chinese).[麻日亮,裴立业,江桦.改进的压缩感知测量矩阵优化方法[J].信号处理,2017,33(2):192-197.]
[8]Malloy ML,Nowak RD.Near-optimal adaptive compressed sensing[J].IEEE Transactions on Information Theory,2014,60(7):4001-4012.
[9]WANG Qiang,LI Jia,SHEN Yi.A survey on deterministic measurement matrix construction algorithms in compressive sensing[J].Acta Electronica Sinica,2013,41(10):2041-2050(in Chinese).[王强,李佳,沈毅.压缩感知中确定性测量矩阵构造算法综述[J].电子学报,2013,41(10):2041-2050.]
[10]LU Cunbo,XIAO Song,QUAN Lei.Construction of compressed sensing measurement matrix based on binary sequence family[J].Journal of Electronics&Information Technology,2016,38(7):1682-1688(in Chinese).[芦存博,肖嵩,权磊.基于二进制序列族的压缩感知测量矩阵构造[J].电子与信息学报,2016,38(7):1682-1688.]
[11]NIU Yakun,YU Zhenming,LI Taoshen,et al.Compressed sampling hard threshold pursuit algorithm with compressive sensing[J].Application Research of Computers,2015,32(8):2286-2288(in Chinese).[牛亚坤,玉振明,李陶深,等.一种压缩采样硬阈值追踪压缩感知重构算法[J].计算机应用研究,2015,32(8):2286-2288.]
[12]Cetin M,Stojanovic I,Onhon O,et al.Sparsity-driven synthetic aperture radar imaging:Reconstruction,autofocusing,moving targets,and compressed sensing[J].IEEE Signal Processing Magazine,2014,31(4):27-40.
[13]Stankovic′S,Orovic′I,Stankovic′L.An automated signal reconstruction method based on analysis of compressive sensed signals in noisy environment[J].Signal Processing,2014,104(13):43-50.
[14]Qureshi MA, Deriche M. A new wavelet based efficient image compression algorithm using compressive sensing[J].Multimedia Tools and Applications,2016,75(12):6737-6754.
[15]ZHAI Xuehan,ZHU Weiping, KANG Bin. Compressed sensing of images combining KSVD and classified sparse representation[J].Computer Engineering and Application,2015,51(6):193-198(in Chinese).[翟雪含,朱卫平,康彬.结合KSVD和分类稀疏表示的图像压缩感知[J].计算机工程与应用,2015,51(6):193-198.]
[2]Metzler C A,Maleki A,Baraniuk R G.From denoising to compressed sensing[J].IEEE Transactions on Information Theory,2016,62(9):5117-5144.
[3]LI Nan,REN Qinghua,SU Yuze.Constructive method of CS measurement matrix in TDCS[J].Computer Engineering and Design,2017,38(1):7-11(in Chinese).[李楠,任清华,苏玉泽.TDCS压缩感知观测矩阵构造方法[J].计算机工程与设计,2017,38(1):7-11.]
[4]Voroninski V,Xu Z.A strong restricted isometry property,with an application to phaseless compressed sensing[J].Applied and Computational Harmonic Analysis,2016,40(2):386-395.
[5]Tillmann A M,Pfetsch M E.The computational complexity of the restricted isometry property,the nullspace property,and related concepts in compressed sensing[J].IEEE Transactions on Information Theory,2014,60(2):1248-1259.
[6]Zhou Nanrun,Zhang Aidi,Zheng Fen,et al.Novel image compression-encryption hybrid algorithm based on key-controlled measurement matrix in compressive sensing[J].Optics&Laser Technology,2014,62(10):152-160.
[7]MA Riliang,PEI Liye,JIANG Hua.Improved optimization algorithm for measurement matrix in compressed sensing[J].Journal of Signal Processing,2017,33(2):192-197(in Chinese).[麻日亮,裴立业,江桦.改进的压缩感知测量矩阵优化方法[J].信号处理,2017,33(2):192-197.]
[8]Malloy ML,Nowak RD.Near-optimal adaptive compressed sensing[J].IEEE Transactions on Information Theory,2014,60(7):4001-4012.
[9]WANG Qiang,LI Jia,SHEN Yi.A survey on deterministic measurement matrix construction algorithms in compressive sensing[J].Acta Electronica Sinica,2013,41(10):2041-2050(in Chinese).[王强,李佳,沈毅.压缩感知中确定性测量矩阵构造算法综述[J].电子学报,2013,41(10):2041-2050.]
[10]LU Cunbo,XIAO Song,QUAN Lei.Construction of compressed sensing measurement matrix based on binary sequence family[J].Journal of Electronics&Information Technology,2016,38(7):1682-1688(in Chinese).[芦存博,肖嵩,权磊.基于二进制序列族的压缩感知测量矩阵构造[J].电子与信息学报,2016,38(7):1682-1688.]
[11]NIU Yakun,YU Zhenming,LI Taoshen,et al.Compressed sampling hard threshold pursuit algorithm with compressive sensing[J].Application Research of Computers,2015,32(8):2286-2288(in Chinese).[牛亚坤,玉振明,李陶深,等.一种压缩采样硬阈值追踪压缩感知重构算法[J].计算机应用研究,2015,32(8):2286-2288.]
[12]Cetin M,Stojanovic I,Onhon O,et al.Sparsity-driven synthetic aperture radar imaging:Reconstruction,autofocusing,moving targets,and compressed sensing[J].IEEE Signal Processing Magazine,2014,31(4):27-40.
[13]Stankovic′S,Orovic′I,Stankovic′L.An automated signal reconstruction method based on analysis of compressive sensed signals in noisy environment[J].Signal Processing,2014,104(13):43-50.
[14]Qureshi MA, Deriche M. A new wavelet based efficient image compression algorithm using compressive sensing[J].Multimedia Tools and Applications,2016,75(12):6737-6754.
[15]ZHAI Xuehan,ZHU Weiping, KANG Bin. Compressed sensing of images combining KSVD and classified sparse representation[J].Computer Engineering and Application,2015,51(6):193-198(in Chinese).[翟雪含,朱卫平,康彬.结合KSVD和分类稀疏表示的图像压缩感知[J].计算机工程与应用,2015,51(6):193-198.]