稀疏恢復(fù)問題中精確恢復(fù)條件的研究
[Abstract]:As a sparse optimization problem in rn space, compressed sensing is aimed at recovering raw data from noisy or partially lost observation data. It has been widely used in signal processing, image denoising, medical imaging and so on. In recent years, compression sensing has been deeply studied and developed rapidly. With the development of modern information technology, the data needed to be stored, processed and analyzed are often large scale, high dimension and complex structure, such as face image, surveillance video, biological information data and so on. Therefore, this paper focuses on sparse optimization problems resulting from the application of compressed sensing to complex high-dimensional data, including sparse solution problems with linear equality and inequality constraints, low-rank matrix restoration problems and low-rank Zhang Liang restoration problems. The relaxation approximation method in compressed sensing is used to solve the NP-hard problem. At present, the algorithm design of relaxation problem has been widely concerned by scholars, but there is not much research on the condition of guaranteeing accurate recovery. In this paper, the exact restoration conditions for the extended sparse optimization problem are systematically studied, and the following results are obtained: 1. Considering the exact restoration condition of sparse solutions of absolute value equations, the equivalent deformation of absolute value equations and bilinear programming is adopted. The problem of solving the sparse solutions of absolute value equations is equivalent to the l0 minimization problem with linear equality and inequality constraints. Based on the analysis of the properties of the range space, the existence and uniqueness conditions of the optimal solution for the convex relaxation of the problem are obtained, and then it is proved that the original problem is equivalent to its convex relaxation under this condition. According to the enlightenment of this research method and the results, we discuss the exact recovery condition of the problem around the sparse optimization problem with linear constraints of general equality and inequality. Some examples are given to verify the correctness of the theoretical results. 2. In this paper, we discuss the exact restoration condition of the low rank matrix restoration problem by non-convex Schatten-p quasi-norm minimization problem, and give a p-RIP condition to guarantee the successful recovery. It is also proved that the p-RIP condition can be encountered with a very high probability by the number of observations. In this paper, three kinds of exact restoration conditions for low rank Zhang Liang restoration problems are generalized to Zhang Liang spaces. Then we consider a low rank Zhang Liang restoration model with both noise-free and noise-free, which is called minimum n- rank approximation, and propose an iterative hard threshold algorithm for solving this problem. It is also proved that the algorithm converges globally linearly at a rate of 1 / 2 under certain conditions for noise-free cases, but the distance between the iterative sequence and the real value in the case of noise decreases rapidly. Numerical experiments verify the theoretical results and show that the algorithm is fast and effective for solving the low nrank Zhang Liang filling problem.
【學(xué)位授予單位】:天津大學(xué)
【學(xué)位級別】:博士
【學(xué)位授予年份】:2016
【分類號】:TP391.41
【相似文獻】
相關(guān)期刊論文 前10條
1 周亮,高翔,朱秀昌;一種新硬閾值算法在臨場感系統(tǒng)中的應(yīng)用[J];計算機工程與應(yīng)用;2005年33期
2 胡智宏;尹小正;路立平;;一維信號小波壓縮的能量動態(tài)自適應(yīng)閾值算法[J];科學(xué)技術(shù)與工程;2012年18期
3 周亮;;一種新型硬閾值算法在圖像去噪中的應(yīng)用[J];軍事通信技術(shù);2005年S1期
4 田玉靜;左紅偉;;小波消噪閾值算法優(yōu)化[J];聲學(xué)技術(shù);2009年04期
5 原玉磊;鄭勇;;一種大視場星圖星點提取的閾值算法[J];海洋測繪;2011年05期
6 李礁;;基于色彩矩陣優(yōu)化的自適應(yīng)閾值算法[J];信息系統(tǒng)工程;2011年08期
7 張?zhí)扈?于鳳芹;;自由分布式FDR假設(shè)檢驗閾值算法的研究[J];武漢理工大學(xué)學(xué)報;2009年06期
8 高翔;周亮;戎舟;;改進型軟閾值算法在臨場感系統(tǒng)中應(yīng)用研究[J];計算機工程與設(shè)計;2006年02期
9 楊海蓉;方紅;張成;韋穗;;基于回溯的迭代硬閾值算法[J];自動化學(xué)報;2011年03期
10 李小靜;李冬梅;梁圣法;;一種改進的迭代硬閾值算法[J];科學(xué)技術(shù)與工程;2014年14期
相關(guān)會議論文 前2條
1 張曉星;彭莉;唐炬;高麗;;一種基于復(fù)小波變換提取PD信號的分塊自適應(yīng)復(fù)閾值算法'[A];08全國電工測試技術(shù)學(xué)術(shù)交流會論文集[C];2008年
2 楊偉;;模糊軟矩陣及其格結(jié)構(gòu)[A];中國運籌學(xué)會模糊信息與模糊工程分會第五屆學(xué)術(shù)年會論文集[C];2010年
相關(guān)博士學(xué)位論文 前4條
1 張敏;稀疏恢復(fù)問題中精確恢復(fù)條件的研究[D];天津大學(xué);2016年
2 賀楊成;半監(jiān)督低秩矩陣學(xué)習及其應(yīng)用[D];上海交通大學(xué);2015年
3 陳梅香;廣義二次矩陣的若干研究[D];福建師范大學(xué);2016年
4 郝曉麗;粒度格矩陣空間模型及其應(yīng)用研究[D];太原理工大學(xué);2009年
相關(guān)碩士學(xué)位論文 前10條
1 田方彥;一種改進的迭代收縮閾值算法[D];河北工業(yè)大學(xué);2015年
2 王玉藏;壓縮感知在無線傳感器網(wǎng)絡(luò)中的應(yīng)用[D];燕山大學(xué);2015年
3 崔翔;基于卷積壓縮感知的確定性測量矩陣研究[D];北京化工大學(xué);2015年
4 吳曼;SDN在IP網(wǎng)絡(luò)的流量調(diào)度應(yīng)用研究[D];電子科技大學(xué);2015年
5 王浩;帶噪聲抑制的流量矩陣估計方法研究[D];電子科技大學(xué);2015年
6 張婷婷;基于低秩矩陣填充與恢復(fù)的圖像去噪方法研究[D];河北工業(yè)大學(xué);2015年
7 鄧愛淘;基于LDPC碼的壓縮感知測量矩陣研究[D];湘潭大學(xué);2015年
8 白平;基于拓展全息矩陣的變胞機構(gòu)創(chuàng)新設(shè)計研究[D];武漢輕工大學(xué);2015年
9 吳越;Vandermonde矩陣的理論與應(yīng)用研究[D];安徽大學(xué);2016年
10 曹萌;幾類Bezout矩陣的研究[D];安徽大學(xué);2016年
,本文編號:2120763
本文鏈接:http://www.wukwdryxk.cn/shoufeilunwen/xxkjbs/2120763.html