【使用L1TV模型分段常数信号的信号去噪/重建】实值或圆值信号的 L^1-TV 正则化的精确算法研究（Matlab代码实现）

 💥💥💞💞欢迎来到本博客❤️❤️💥💥

🏆博主优势：🌞🌞🌞博客内容尽量做到思维缜密，逻辑清晰，为了方便读者。

⛳️座右铭：行百里者，半于九十。

📋📋📋本文目录如下：🎁🎁🎁

目录

💥1 概述

📚2 运行结果

2.1 算法1

2.2 算法2 

🎉3 参考文献

🌈4 Matlab代码及详细文章

---

💥1 概述

文献来源：

我们考虑实值线或单位圆上单变量信号的L1-TV正则化。虽然实数数据空间导致一个凸优化问题，但对于圆值数据，问题是非凸的。在本文中，我们推导了两个数据空间的精确算法。一个关键的组成部分是将无限搜索空间减少到有限的配置集，可以通过维特比算法进行扫描。为了降低所涉及的表格计算复杂性，我们将距离变换技术扩展到非均匀网格和循环数据空间。总体而言，所提出的算法的复杂性为O(KN)，其中N是信号的长度，K是数据集中不同值的数量。特别地，对于量化数据，复杂性为O(N)。这是第一个用于具有圆值数据的TV正则化的精确算法，并且在假设后者被量化的情况下，与标量数据的最先进方法相竞争。关键词：总变化正则化、总循环变化、圆值数据、最小绝对偏差、动态规划、距离变换。

原文摘要：

We consider L 1 -TV regularization of univariate signals with values on the real line or on the unit circle. While the real data space leads to a convex optimization problem, the problem is non-convex for circle-valued data. In this paper, we derive exact algorithms for both data spaces. A key ingredient is the reduction of the infinite search spaces to a finite set of configurations, which can be scanned by the Viterbi algorithm. To reduce the computational complexity of the involved tabulations, we extend the technique of distance transforms to non-uniform grids and to the circular data space. In total, the proposed algorithms have complexity O(KN) where N is the length of the signal and K is the number of different values in the data set. In particular, the complexity is O(N) for quantized data. It is the first exact algorithm for TV regularization with circle-valued data, and it is competitive with the state-of-the-art methods for scalar data, assuming that the latter are quantized.

Keywords: Total variation regularization, total cyclic variation, circle-valued data, least absolute deviations, dynamic programming, distance transform 

📚2 运行结果

2.1 算法1

2.2 算法2 

 

 

 

 部分代码：

%%% Demo for denoising a circle-valued signal by the L1-TV model

% create random signal (smoothed pcw constant signal)
rng(12345) % random seed for reproducibility
N = 2000;
t = 2*pi;
lambda = 20 / N;
K = 20;
sigma= 0.3;
innovation = randCP((rand([N, 1])-0.5) * t, lambda );
signalUnwrapped = cumsum(innovation);
h = fspecial('Gaussian', [N/10, 1], 10);
smoothed = conv(signalUnwrapped, h, 'same');
groundTruth = wrapAngle(smoothed);

% add noise
y = wrapAngle(groundTruth + sigma* randl(size(groundTruth)));

% perform restoration using L1TV_Circ
alpha = sqrt(N)*sigma;
x = L1TV_Circ(y, alpha);

% plot the results
figure('Color', 'w')
subplot(1,2,1)
plot(y, '.')
ylim([-pi,pi])
title('Data with values on the unit circle')

subplot(1,2,2)
plot(x, '.')
ylim([-pi,pi])
title(sprintf('L1TV restoration, SNR improvement: %.2f dB', deltaSNR(groundTruth, y, x, 'circ')))

🎉3 参考文献

部分理论来源于网络，如有侵权请联系删除。

[1]M. Storath, A. Weinmann, M. Unser. Exact algorithms for L^1-TV regularization of real-valued or circle-valued signals. SIAM Journal on Scientific Computing, 38(1), A614-A630, 2016

🌈4 Matlab代码及详细文章