人工智能培训

搜索

人工智能论文:一种规则化图像重建的在线即插即用算法(An Online Plug-and-Play Algorithm for Regularized Imag

[复制链接]
liwei906666 发表于 2018-9-14 09:03:24 | 显示全部楼层 |阅读模式
liwei906666 2018-9-14 09:03:24 211 0 显示全部楼层
人工智能论文:一种规则化图像重建的在线即插即用算法(An Online Plug-and-Play Algorithm for Regularized Image Reconstruction)即插即用先验(PnP)是一种强大的框架,可通过在迭代算法中使用高级降噪器来规范成像反问题。最近的实验证据表明,PnP算法在一系列成像应用中实现了最先进的性能。在本文中,我们介绍了一种基于迭代收缩/阈值算法(ISTA)的新的在线PnP算法。所提出的算法在每次迭代时仅使用测量的子集,这使得它可以扩展到非常大的数据集。我们为PnP-ISTA的批量和在线变体提供了一种新的理论收敛性分析,用于不一定与近端算子相对应的降噪器。我们还提供了模拟,说明了算法在衍射图像中的图像重建的适用性。本文的结果有可能将PnP框架的适用范围扩展到非常大且冗余的数据集。
Plug-and-play priors (PnP) is a powerful framework for regularizing imaginginverse problems by using advanced denoisers within an iterative algorithm.Recent experimental evidence suggests that PnP algorithms achievestate-of-the-art performance in a range of imaging applications.In this paper,we introduce a new online PnP algorithm based on the iterativeshrinkage/thresholding algorithm (ISTA).The proposed algorithm uses only asubset of measurements at every iteration, which makes it scalable to verylarge datasets.We present a new theoretical convergence analysis, for bothbatch and online variants of PnP-ISTA, for denoisers that do not necessarilycorrespond to proximal operators.We also present simulations illustrating theapplicability of the algorithm to image reconstruction in diffractiontomography.The results in this paper have the potential to expand theapplicability of the PnP framework to very large and redundant datasets.人工智能论文:一种规则化图像重建的在线即插即用算法(An Online Plug-and-Play Algorithm for Regularized Image Reconstruction) FpS65X44Iz65sImT.jpg
URL地址:https://arxiv.org/abs/1809.04693     ----pdf下载地址:https://arxiv.org/pdf/1809.04693    ----人工智能论文:一种规则化图像重建的在线即插即用算法(An Online Plug-and-Play Algorithm for Regularized Image Reconstruction)
回复

使用道具 举报

您需要登录后才可以回帖 登录 | 立即注册

本版积分规则 返回列表 发新帖

liwei906666当前离线
新手上路

查看:211 | 回复:0

快速回复 返回顶部 返回列表