基于频域稀疏压缩感知的星载SAR稀疏重航过3维成像

星载合成孔径雷达(SAR)稀疏重航过3维成像技术通过交轨向的多次飞行观测,获得观测场景的第3维分辨。该文给出了单颗卫星SAR稀疏重航过轨道分布,为有效缩短重访时间,同时给出了编队双星SAR轨道分布,对应的交轨向等效孔径长度为20 km。提出了一种基于干涉处理和频域压缩感知(CS)的稀疏3维成像方法,利用稀疏重航过中的部分回波形成参考3维复图像,对待重建SAR 3维图像信号进行干涉处理,使信号在频域具备稀疏性。在大轨道分布范围下,建立频域距离向-交轨向线性测量矩阵,利用CS理论联合求解稀疏表征下的图像频谱,避免交轨向和距离向的回波信号耦合。将求解所得频谱逆变换至空间域,可得到观测场景的3维图像重建结果。仿真结果表明,该文方法在稀疏采样率74.4%条件下,仍可获得与满采样成像性能相当的结果,验证了干涉处理频域稀疏方法在星载SAR 3维成像中的有效性。     

基于概率稀疏随机矩阵的压缩数据收集方法

测量矩阵设计是应用压缩感知理论解决实际问题的关键。该文针对无线传感器网络压缩数据收集问题设计了一种概率稀疏随机矩阵。该矩阵可在减少参与投影值计算节点个数的同时,让参与投影值计算的节点分布集中化,从而降低数据收集的通信能耗。在此基础上,为提高网络数据重构精度,又提出一种适用于概率稀疏随机矩阵优化的测量矩阵优化算法。仿真实验结果表明,与稀疏随机矩阵和稀疏Toeplitz测量矩阵相比,采用优化的概率稀疏随机矩阵作为压缩数据收集的测量矩阵可显著降低通信能耗,且重构误差更小。     

基于压缩感知和红外玫瑰线扫描的红外图像重构

红外亚成像制导技术是由点源探测技术到成像制导技术的一种过渡, 由单元探测器和光机扫描装置组成.红外玫瑰线扫描亚成像系统是亚成像制导中的一种, 红外玫瑰线扫描亚成像系统按照特定的图案采集视场中的部分数据并得到一幅含有目标位置信息的亚图像.受单像素相机的启发, 主要研究红外玫瑰线扫描亚成像系统中的压缩成像.压缩感知可以在更少的采样数据条件下重构红外图像, 其应用到红外亚成像制导系统中一个关键的问题就是观测矩阵的构造.关于随机观测矩阵的研究已经比较广泛, 但随机矩阵很难实现.本文提出了一种简单的适用于红外玫瑰线扫描亚成像系统的确定性观测矩阵.此外还提出了一种快速有效的恢复算法, 称为优化子空间追踪算法.仿真结果显示构造的观测矩阵能够压缩和重构红外图像, 且重构效果优于随机高斯观测矩阵和随机伯努利观测矩阵, 提出的恢复算法也具有较好的表现.     

基于随机滤波的探地雷达成像方法

探地雷达是一种超宽带雷达系统,若按传统的奈奎斯特采样,雷达回波信号需要大量空间存储。压缩感知可以实现利用少量的测量值对稀疏信号进行重构,其中最为关键的是测量矩阵和重构算法的选择。本文将压缩感知应用于探地雷达成像,并利用随机滤波的思想选择测量矩阵,可以有效减少测量矩阵中非零值的个数。利用正交匹配追踪算法对信号进行重构,算法简单,降低了数据的存储量和运算复杂度,该算法同样可以对时间和空间上同时压缩的数据进行成像。最后,本文给出基于时间连续信号的GPR接收机一种CS实现方案。仿真结果表明,本文提出的成像方法可以以少量数据精确地对信号进行重构,并且运算量少。      

Compressed Blind Signal Reconstruction Model and Algorithm

The model of inherent connection between underdetermined blind signal separation and compressed sensing (CS) is analyzed first; then, the mathematical model of underdetermined blind signal reconstruction is built using CS. More specifically, the mixing matrix is estimated by exploiting the wavelet packet transform and k-means clustering methods up to permutation and scaling indeterminacy, and then, the measurement matrix and the measurement equation are obtained. To reconstruct the underdetermined sparse source signals, the proposed semi-blind compressed reconstruction algorithm is derived based on the blind signal reconstruction model and compressive sampling matching pursuit (CoSaMP) method. Our simulation results demonstrate that the proposed scheme is effective, irrespective of artificial data or real data. Moreover, the proposed scheme can be adjusted for different applications by modifying the mixing matrix estimation method and CoSaMP method with respect to the correspondence conditions.     

半张量积低存储压缩感知方法研究

由于随机观测矩阵的随机性,存在数据存储量大、内存占用率高、数据计算量大以及难以面向大规模实际应用等问题.为此,提出了一种可有效降低随机观测矩阵所占存储空间的半张量积压缩感知（STP-CS）方法.利用该方法,构建低维随机观测矩阵,经奇异值分解（SVD）优化后对原始信号进行采样,并利用拟合0-范数的迭代重加权方法进行重构.实验利用2维灰度图像进行测试,并对重构图像的峰值信噪比,结构相似度等指标进行了统计和比较.实验结果表明,本文所述的STP-CS方法在不改变随机观测矩阵数据类型的前提下,可将观测矩阵减小至传统CS模型中观测矩阵所占内存空间的1/256（甚至更低）,同时仍保持很高的重构质量.     

One-bit compressed sensing recovery algorithm robust to perturbation

One-bit compressed sensing(CS) technology reconstructs the sparse signal when the available measurements are reduced to only their sign-bit. It is well known that CS reconstruction should know the measurement matrix exactly to obtain a correct result. However, the measurement matrix is probably perturbed in many practical scenarios. An iterative algorithm called perturbed binary iterative hard thresholding (PBIHT) is proposed to reconstruct the sparse signal from the binary measurements (sign measurements) where the measurement matrix experiences a general perturbation. The proposed algorithm can reconstruct the original data without any prior knowledge about the perturbation. Specifically, using the ideas of the gradient descent, PBIHT iteratively estimates signal and perturbation until the estimation converges. Simulation results demonstrate that, under certain conditions, PBIHT improves the performance of signal reconstruction in the perturbation scenario.     

基于稀疏贝叶斯学习的CT图像重构

计算机断层成像是医学检查的常用方法,但是检查中过量的辐射可能对病人造成二次伤害。基于此提出了一种稀疏贝叶斯学习(Sparse Bayesian Learning,SBL)的肺部计算机断层成像(Computed Tomography,CT)图像重构方法,首先应用高斯随机分布矩阵对肺部图像进行测量,并建立基于小波变换的稀疏字典,最后应用稀疏贝叶斯学习算法对图像进行重构。仿真实验结果显示,该方法能够实现对肺部图像重构,当压缩率为0.6的时候,重构的肺部组织图像的峰值信噪比达到34.1809,满足医学检查的需求。该方法能够降低医学检查中的辐射伤害,具有重要的理论研究意义和实际应用价值。     

Fast reconstruction method for compressed sensing model with semi-tensor product

To reduce the storage space of random measurement matrix and improve the reconstruction efficiency for compressed sensing (CS),a new sampling approach for CS with semi-tensor product (STP-CS) was proposed.The proposed approach generated a low dimensional random measurement matrix to sample the sparse signals.Then the solutions of the sparse vector were estimated group by group with a l_q-minimization (0$\frac{M}{t}\text{×}\frac{N}{t}$ and decrease tow orders of magnitude of time that for conventional CS,while maintaining the reconstruction quality.Numerical results also show that the reconstruction time can be effectively improved 260 for the image size of 1 024×1 024.     

非对称代价函数的稀疏卷积非负矩阵分解方法

提出一种基于非对称代价函数的稀疏卷积非负矩阵分解方法。该方法利用板仓-斋藤距离作为目标代价函数来衡量目标矩阵与重建矩阵的差异,使得较小的矩阵元素具有较小的重建误差,并且该代价函数具有尺度不变性的特点。为了考察其在弱语音成分重建方面的优势,将本文提出的算法应用于耳语音谱分解及重建实验。实验结果表明,与基于欧氏距离和基于Kullback-Leibler(K-L)散度的卷积非负矩阵分解算法相比,本文算法对于弱语音成分具有更好的重构效果,重建后的语音信号具有较大的可懂度。      

一种基于特征值分解的测量矩阵优化方法

测量矩阵是压缩感知中一个很重要的部分,为了减小测量矩阵与稀疏变换矩阵的互相干性,从而改善重建质量,本文首先通过测量矩阵和稀疏变换矩阵的乘积构造得到一个Gram矩阵,然后定义了一种基于Gram矩阵非对角线元素的整体互相干系数,推导出整体互相干系数与Gram矩阵特征值之间的关系。在此基础上,我们提出了一个最优化模型,在不改变Gram矩阵特征值和的前提下,让每个大于零的特征值的大小都为它们和的平均值,使得测量矩阵和稀疏变换矩阵的整体互相干系数达到最小,从而优化了测量矩阵的性能。将该方法用在一些已知的测量矩阵上,实验结果中矩阵的优化速度快,并且用优化矩阵所得的图像的PSNR有所提高,表明本文优化测量矩阵的方法在重建效果和优化速度方面都有一定的优势。      

Image representation by compressive sensing for visual sensor networks

This paper addresses the image representation problem in visual sensor networks. We propose a new image representation method for visual sensor networks based on compressive sensing (CS). CS is a new sampling method for sparse signals, which is able to compress the input data in the sampling process. Combining both signal sampling and data compression, CS is more capable of image representation for reducing the computation complexity in image/video encoder in visual sensor networks where computation resource is extremely limited. Since CS is more efficient for sparse signals, in our scheme, the input image is firstly decomposed into two components, i.e., dense and sparse components; then the dense component is encoded by the traditional approach (JPEG or JPEG 2000) while the sparse component is encoded by a CS technique. In order to improve the rate distortion performance, we leverage the strong correlation between dense and sparse components by using a piecewise autoregressive model to construct a prediction of the sparse component from the corresponding dense component. Given the measurements and the prediction of the sparse component as initial guess, we use projection onto convex set (POCS) to reconstruct the sparse component. Our method considerably reduces the number of random measurements needed for CS reconstruction and the decoding computational complexity, compared to the existing CS methods. In addition, our experimental results show that our method may achieves up to 2 dB gain in PSNR over the existing CS based schemes, for the same number of measurements.     

基于图稀疏正则化多测量向量模型的高光谱压缩感知重建

压缩感知重建是解决高光谱现有成像模式数据量大冗余度高问题的一个有效机制。针对高光谱图像的多通道特性,该文建立了高光谱压缩感知的多测量向量模型,编码端使用随机卷积算子对各通道进行快速采样,生成测量向量矩阵。解码端构建图稀疏正则化的联合重建模型,在稀疏变换域将高光谱图像分解为谱间的关联成分和差异成分,通过图结构化稀疏度量表征关联成分的空谱相关性,并约束谱间差异成分的稀疏性。进一步提出模型求解的交替方向乘子迭代算法,通过引入辅助变量与线性化技巧,使得每一子问题均存在解析解,降低了模型求解的复杂度。对多个实测数据集进行了对比实验,实验结果验证了该文模型与算法的有效性。     

稀疏随机矩阵有限等距性质分析

稀疏随机矩阵由于具有存储容量小、编码和重构复杂度低、易于更新等优良特性而适用于分布式应用。为确保稀疏随机矩阵可作为压缩感知观测矩阵,该文证明了稀疏随机矩阵的有限等距性质(RIP)。首先,证明了测量矩阵满足有限等距性质等价于其子矩阵的格拉姆矩阵特征值分布于1附近;在此基础上,证明了当测量值个数满足特定条件时,稀疏随机矩阵以接近于1的概率满足有限等距性质。仿真实验表明,稀疏随机矩阵在保证稀疏信号精确重建的同时,大大节约了测量和重建所需的时间。     

基于最优观测矩阵的压缩信道感知

信道估计技术作为获得信道衰落信息的方法,是提高无线信道传输接收性能的关键技术。而物理多径信道固有的稀疏性,使得将压缩感知（CS）理论用于稀疏多径信道的估计成为可能。由于传统的线性估计方法没有考虑信道的固有稀疏性,因而在训练序列数目较少的情况下,压缩信道估计的重构效果要明显优于传统的最小二乘估计方法,在获得同样估计性能的情况下,需要的训练序列长度也大大减少,提高了频谱资源利用率,体现了压缩信道估计出色的估计性能。本文在应用CS理论进行稀疏信道估计的过程中,通过减小观测矩阵的列向量相关性,产生最优观测矩阵的方法,从而让压缩信道估计的性能得到进一步的改善。      

Deterministic Sensing Matrices Based on Multidimensional Pseudo-Random Sequences

An approach is proposed for producing compressed sensing (CS) matrices via multidimensional pseudo-random sequences. The columns of these matrices are binary Gold code vectors where zeros are replaced by ?1. This technique is mainly applied to restore sub-Nyquist-sampled sparse signals, especially image reconstruction using block CS. First, for the specific requirements of message length and compression ratio, a set Λ which includes all preferred pairs of m-sequences is obtained by a searching algorithm. Then a sensing matrix A _M×N is produced by using structured hardware circuits. In order to better characterize the correlation between any two columns of A, the average coherence is defined and the restricted isometry property (RIP) condition is described accordingly. This RIP condition has strong adaptability to different sparse signals. The experimental results show that with constant values of N and M, the sparsity bound of A is higher than that of a random matrix. Also, the recovery probability may have a maximum increase of 20 % in a noisy environment.     

An adaptive measurement scheme based on compressed sensing for wideband spectrum detection in cognitive WSN

An Adaptive Measurement Scheme (AMS) is investigated with Compressed Sensing (CS) theory in Cognitive Wireless Sensor Network (C-WSN). Local sensing information is collected via energy detection with Analog-to-Information Converter (AIC) at massive cognitive sensors, and sparse representation is considered with the exploration of spatial temporal correlation structure of detected signals. Adaptive measurement matrix is designed in AMS, which is based on maximum energy subset selection. Energy subset is calculated with sparse transformation of sensing information, and maximum energy subset is selected as the row vector of adaptive measurement matrix. In addition, the measurement matrix is constructed by orthogonalization of those selected row vectors, which also satisfies the Restricted Isometry Property (RIP) in CS theory. Orthogonal Matching Pursuit (OMP) reconstruction algorithm is implemented at sink node to recover original information. Simulation results are performed with the comparison of Random Measurement Scheme (RMS). It is revealed that, signal reconstruction effect based on AMS is superior to conventional RMS Gaussian measurement. Moreover, AMS has better detection performance than RMS at lower compression rate region, and it is suitable for large-scale C-WSN wideband spectrum sensing.     

联合感知矩阵优化的穿墙MIMO阵列稀疏成像方法

采用压缩感知理论的穿墙稀疏成像恢复算法需要感知矩阵满足有限等距性质(RIP),最直接的验证方法是判定感知矩阵的相干系数是否较小。针对现有的穿墙多输入多输出(MIMO)阵列稀疏成像方法没有验证感知矩阵是否满足RIP 性质而容易出现重构失败并导致成像模糊的问题,提出一种联合感知矩阵优化的穿墙MIMO阵列稀疏成像方法。该方法首先依据配置指标将阵元配置为两端发中间收和分时复用的模式,既能使感知矩阵的相干系数最小,又能获得均匀而不冗余的等效虚拟阵元;然后,从中选取部分能够满足感知矩阵相干系数最小的虚拟阵元组合,使用可分离逼近结构稀疏恢复算法充分考虑扩展目标信号的结构稀疏先验信息对其进行稀疏成像重构;最后,选取成像性能指标较好的一组作为成像结果。仿真和实验结果表明,该方法降低了运算量和虚拟阵元间的干扰,节约了硬件成本,提高了算法的稀疏重构性能,获得了高分辨的穿墙扩展目标成像。     

Image compressed sensing based on wavelet transform in contourlet domain

Compressed sensing (CS) has been widely concerned and sparsity of a signal plays a crucial role in CS to exactly recover signals. Contourlet transform provides sparse representations for images, so an algorithm of CS reconstruction based on contourlet is considered. Meanwhile, taking into account the computation and the storage of large random measurement matrices in the CS framework, we are trying to introduce the wavelet transform into the contourlet domain to reduce the size of random measurement matrices. Several numerical experiments demonstrate that this idea is feasible. The proposed algorithm possesses the following advantages: reduced size of random measurement matrix and improved recovered performance.     

基于DWT-IRLS算法的超声断层成像方法研究

超声成像因非侵入式、成本低且实时性好而被广泛应用。超声系统需要大量的采集通道数据和较高的采样率来提高图像重建质量,导致成像耗时,系统复杂。压缩感知(compressed sensing, CS)算法能够在欠采样的条件下用较少的测量值重构出原始信号。因此,针对系统面临的采样率高,数据量大的问题,本文将CS理论中的DWT-IRLS算法应用在超声成像中,通过离散小波变换基(discrete wavelet transformation, DWT)对超声数据进行稀疏转换,对高低频系数进行采样测量,并使用迭代重加权最小二乘法(iterative reweighted least squares, IRLS)进行测量系数重构,最后对变换域系数进行DWT逆转换得到重建图像。通过实验分析,以50%原始数据重建图像效果逐渐趋于稳定,在均方误差和峰值信噪比方面进行对比分析,DWT-IRLS算法相比较于DWT-OMP、DWT-CoSamp和DCT-IRLS等重构算法,成像质量更高,细节特征更为明显。