压缩感知理论在蝗灾监测中的应用

非洲蝗虫灾害给当地农作物带来了巨大的损害,为了实时监测蝗虫迁徙动向和蝗虫变异状态,提出一种改进的压缩采样匹配追踪(uCoSaMP)蝗虫图像重构算法.针对观测矩阵性能不佳的现状,通过构造新观测矩阵,有效提高观测值的重构潜能.在原子选择阶段,利用修正因子自适应地修正残差与观测矩阵各列的相关性,提高候选集的正确率;在原子剔除...     

压缩感知重构匹配类算法分析

压缩感知理论是一种利用信号的稀疏性或可压缩性而把采样与压缩融为一体的新理论体系,它成功地克服了传统理论中采样数据量大、资源浪费严重等问题。该理论的研究方向主要包括信号的稀疏表示、测量矩阵的设计和信号的重构算法。其中信号的重构算法是该理论中的关键部分,也是近年来研究的热点。本文主要对匹配追踪类重构算法作了详细介绍,并通过仿真实验结果对这些算法进行了对比和分析。     

压缩感知综述

压缩感知理论的诞生使得采样速率与信号的结构和内容相关,并以低于奈奎斯特采样定理要求的频率采样、编码和重构。在实际应用中,为解决数据冗余和资源浪费的瓶颈问题开拓了一条新道路,也为其他学科发展提供了新的契机。从发展历史和研究现状等方面入手,对稀疏表示、测量矩阵的构造、稀疏重构算法和主要应用方面进行了详细的梳理和研究。对当前研究的热点、难点作了分析和探讨,并指出了未来的发展方向和应用前景。     

单比特压缩感知帧定时同步

相对于非压缩感知帧定时同步方法,压缩感知帧定时同步方法可以降低系统的能量消耗,降低模拟数字转换器的设计难度。相对于压缩感知技术,单比特压缩感知仅保留观测值的符号信息,可进一步降低系统的能量消耗,降低模拟数字转换器的设计难度。为此,将单比特压缩感知技术引入到帧定时同步中,提出了一种基于单比特压缩感知的帧定时同步方法。提出方法首先在帧定时变换域对接收信号进行单比特的压缩采样;随后,利用采样到的比特流重构出用于帧定时同步的定时度量;最后,根据相关法帧定时同步准则,搜索重构到的定时度量,找到帧定时同步的索引位置。分析与仿真结果表明,相对于压缩感知帧定时同步方法,在相同的比特开销情况下,提出方法可改善帧定时同步的正确同步概率;在相同的正确同步概率情况下,提出方法所需比特数更少。同时,提出方法的量化过程仅需要电平比较器,降低了模拟数字转换器设计难度。     

基于压缩感知CoSaMP算法的精确重构

为有效解决压缩采样匹配追踪(Compressive Sampling Matching Pursuit, CoSaMP)算法对稀疏度K值的依赖问题,提高重构精度,提出了一种根据峰值信噪比增减变化趋势来确定最佳迭代次数的CoSaMP改进算法。先将PSNR算式进行数学推导演变,将算式中未知的原始信号巧妙转换为已知信号,并证明了此转换式与PSNR算式有相同增减性,在迭代过程中基于此转换式可根据各列稀疏度的不同,自适应的确定不同列的最佳迭代次数,从而保证更高的重构精度。理论分析和实验仿真表明,改进的CoSaMP算法比原有算法有更理想的重构效果,与其它重构算法相比有更高的重构成功率,并且更具高效性和实用性。     

压缩感知中贪婪重构算法研究

压缩感知理论是利用信号的稀疏性,通过少量的观测值就可以实现对该信号的精确重构。贪婪类算法是压缩感知重构步骤中广泛应用的一类算法。该文主要对该类算法中典型的三种算法在存在噪声环境中进行了综合分析比较。首先从理论方面分析了三种算法,给出了实现过程;然后在不同稀疏度情况下,对三种贪婪算法重构性能进行综合比较。根据理论分析结果和仿真结果,得出相应的结论。     

压缩感知中贪婪重构算法研究

压缩感知理论是利用信号的稀疏性,通过少量的观测值就可以实现对该信号的精确重构。贪婪类算法是压缩感知重构步骤中广泛应用的一类算法。该文主要对该类算法中典型的三种算法在存在噪声环境中进行了综合分析比较。首先从理论方面分析了三种算法,给出了实现过程;然后在不同稀疏度情况下,对三种贪婪算法重构性能进行综合比较。根据理论分析结果和仿真结果,得出相应的结论。     

基于压缩感知的信道估计技术

通信系统中的传统信道估计方法均基于多信道密集型假设,导致频谱利用率低下,压缩感知理论为解决这一问题提供了一种新的途径。本文介绍了压缩感知基本理论,探讨了压缩感知应用于信道估计的可行性,详细分析了压缩感知信道估计技术的MP算法、OMP算法、CoSaMP等几种重构算法。研究表明基于压缩感知理论的信道估计方法能利用较少的导频信号达到与传统方法相比拟的估计性能,从而提高频谱利用率。     

压缩感知在医学图像压缩中的应用

调研压缩感知的数学理论基础和常用方法,包括稀疏变换、测量矩阵和重构算法,利用Matlab软件实现压缩感知实验,比较几种测量矩阵的性能,提出双阈值分块正交匹配追踪重构算法。根据图像不同区域信息量的不同,采取分块处理的方法并加入采样阈值,针对不同子图像块采取不同采样率,提高采样效率;加入判断阈值,降低重构效果对采样阈值的依赖。实验结果表明,该方法能够以较低的采样率实现较高的重构精度,使压缩感知在医学图像压缩方面得到了较好应用。     

基于贝叶斯压缩感知的信号重构

本文提出了基于贝叶斯压缩感知的信号重构算法,将压缩感知理论应用于信号的压缩传输以及重构,该算法将压缩感知问题转化为线性回归问题,逐步推演出结果向量之间的迭代关系,最后通过迭代以得到原始信号的精确重构. 仿真说明了贝叶斯压缩感知在信号处理中的应用,结果表明该算法对一维和二维信号的压缩重构有很好的效果.     

A survey on distributed compressed sensing: theory and applications

The compressed sensing (CS) theory makes sample rate relate to signal structure and content. CS samples and compresses the signal with far below Nyquist sampling frequency simultaneously. However, CS only considers the intra-signal correlations, without taking the correlations of the multi-signals into account. Distributed compressed sensing (DCS) is an extension of CS that takes advantage of both the inter- and intra-signal correlations, which is wildly used as a powerful method for the multi-signals sensing and compression in many fields. In this paper, the characteristics and related works of DCS are reviewed. The framework of DCS is introduced. As DCS’s main portions, sparse representation, measurement matrix selection, and joint reconstruction are classified and summarized. The applications of DCS are also categorized and discussed. Finally, the conclusion remarks and the further research works are provided.     

A survey on one-bit compressed sensing: theory and applications

In the past few decades, with the growing popularity of compressed sensing (CS) in the signal processing field, the quantization step in CS has received significant attention. Current research generally considers multi-bit quantization. For systems employing quantization with a sufficient number of bits, a sparse signal can be reliably recovered using various CS reconstruction algorithms.Recently, many researchers have begun studying the one-bit quantization case for CS. As an extreme case of CS, one-bit CS preserves only the sign information of measurements, which reduces storage costs and hardware complexity. By treating one-bit measurements as sign constraints, it has been shown that sparse signals can be recovered using certain reconstruction algorithms with a high probability. Based on the merits of one-bit CS, it has been widely applied to many fields, such as radar, source location, spectrum sensing, and wireless sensing network.In this paper, the characteristics of one-bit CS and related works are reviewed. First, the framework of one-bit CS is introduced. Next, we summarize existing reconstruction algorithms. Additionally, some extensions and practical applications of one-bit CS are categorized and discussed. Finally, our conclusions and the further research topics are summarized.     

压缩感知在无线传感网络的应用综述

随着信息技术的发展,近些年压缩感知技术格外引人瞩目,在图像视频编码、雷达及微波辐射成像、气象卫星、图像加密、物联网等领域展现出强大的功能与发展前景。首先介绍了压缩感知在无线传感网络领域的发展及研究现状,然后从压缩感知仿真实验和实例、压缩感知的测量方案、压缩感知的解压缩方案、压缩感知在无线传感网络的具体应用四个方面阐明了压缩感知在无线传感网络领域的优势,最后对压缩感知的前景进行了展望。     

压缩感知基本理论:回顾与展望

随着信息社会的迅速发展,人们对数字信息的需求越来越大。同时,人们对信号的采样速率、传输速度和存储空间的要求也变得越来越高。如何在保持信号信息的同时尽可能地减少信号的采样数量?Candès在2006年的国际数学家大会上介绍了一种称为压缩感知的新颖信号采样理论,指出:只要远少于传统Nyquist采样定理所要求的采样数即可精确或高概率精确重建原始信号。围绕压缩感知的稀疏字典设计、测量矩阵设计、重建算法设计这3个核心问题,对其基本理论和主要方法进行了系统阐述,同时指出了压缩感知有待解决的若干理论问题与关键技术。     

基于压缩感知理论的图像融合

像素级图像融合需要对待融合图像的像素灰度进行综合分析与处理,而巨大的数据量给图像融合带来了极大的不便。压缩感知理论的兴起提供了一种新的数据采集和压缩方法 ,它利用图像的部分信息重构原始图像,有效减轻了图像处理的计算复杂度。本文基于对压缩感知理论的研究,把压缩感知的方法应用于图像融合,并尝试通过对压缩感知观测矩阵的改进提高融合质量。文中设计出一种适应于DCT域的射线采样矩阵,通过与随机生成的观测矩阵融合效果的比较,证明该方法取得了良好的融合效果。     

一种基于压缩感知理论的纹理分类方法

针对传统纹理分类方法计算复杂的问题,本文基于bag-of-words模型提出了一种简单、新奇的纹理分类方法。在特征提取阶段,使用NSCT滤波器对局部图像块进行映射投影,然后通过观测矩阵提取其随机测量值特征;在纹理分类阶段,直接将随机特征嵌入到bag-of-words环境,并且直接在压缩域内进行学习和分类。利用纹理图像的稀疏性,本文提出的特征提取方法简单,并且在性能和复杂度上都优于传统特征提取方法。最后使用CUReT数据库进行数值试验,并与patch、patch-MRF、MR8、LBP四种最经典的方法进行比对,本文方法在分类精度以及实时性上有重要的改进。     

启发式重构算法在压缩传感中的应用研究

针对传统算法在处理传感器网络的大规模信号时,运算复杂度显著增大,性能急剧下降的问题,提出了启发式同步自适应迭代阈值重构算法。采用启发式差错控制函数选择代价最少的方向逐行同步收缩逼近最优解,并结合由自适应递减幂指数参数所确定的非线性阈值函数,进一步判断修正重构信号。仿真结果表明,启发式同步自适应迭代阈值重构算法以更少的测量值和迭代次数重构信号,其信噪比提高了60 dB。     

Compressive rendering: a rendering application of compressed sensing

Recently, there has been growing interest in compressed sensing (CS), the new theory that shows how a small set of linear measurements can be used to reconstruct a signal if it is sparse in a transform domain. Although CS has been applied to many problems in other fields, in computer graphics, it has only been used so far to accelerate the acquisition of light transport. In this paper, we propose a novel application of compressed sensing by using it to accelerate ray-traced rendering in a manner that exploits the sparsity of the final image in the wavelet basis. To do this, we raytrace only a subset of the pixel samples in the spatial domain and use a simple, greedy CS-based algorithm to estimate the wavelet transform of the image during rendering. Since the energy of the image is concentrated more compactly in the wavelet domain, less samples are required for a result of given quality than with conventional spatial-domain rendering. By taking the inverse wavelet transform of the result, we compute an accurate reconstruction of the desired final image. Our results show that our framework can achieve high-quality images with approximately 75 percent of the pixel samples using a nonadaptive sampling scheme. In addition, we also perform better than other algorithms that might be used to fill in the missing pixel data, such as interpolation or inpainting. Furthermore, since the algorithm works in image space, it is completely independent of scene complexity.     

压缩感知理论在语音通信系统中的应用研究

本文介绍了压缩感知理论,并把这一理论应用在语音通信系统中。根据语音信号的短时平稳性、频域的可稀疏性,采用结合DCT和FFT两种不同稀疏基的OMP重构算法对语音信号进行重构,并对重构结果进行主客观评价。实验结果表明:固定帧长和压缩感知采样率M/N对语音信号重构性能有影响;以及当M/N>0.4时,DCT稀疏基表现出更好的重构性能;当M/N<0.4时,FFT稀疏基表现出更好的重构性能。     

Robust 1-bit compressive sensing via variational Bayesian algorithm

In a compressive sensing (CS) framework, a sparse signal can be stably reconstructed at a reduced sampling rate. Quantization and noise corruption are inevitable in practical applications. Recent studies have shown that using only the sign information of measurements can achieve accurate signal reconstruction in a CS framework. We consider the problem of reconstructing a sparse signal from 1-bit quantized, Gaussian noise corrupted measurements. In this paper, we present a variational Bayesian inference based 1-bit compressive sensing algorithm, which essentially models the effect of quantization as well as the Gaussian noise. A variational message passing method is adopted to achieve the inference. Through numerical experiments, we demonstrate that our algorithm outperforms state-of-the-art 1-bit compressive sensing algorithms in the presence of Gaussian noise corruption.