Scholarlink

<p>In this work, we improve the results of various partial differential equation (PDE) models, which are used in image denoising, applying reservoir computing networks (RCNs). RCN is a kind of recurrent neural network. Echo state neural network (ESN), which is one type of RCNs, has better quality of denoising and RCNs possess the superiority of shorter training time compared to deep neural networks like convolutional neural network (CNN). The considered PDE models are blending models and consist of isotropic diffusion (ID), Perona-Malik (PM) and some other PDE models, which are applicable in image denoising. In experimental results, the efficiency of using ESN on the output image, which is denoised by blending PDE model, is presented using PSNR criteria comparing with using the PDE model singly.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Application of reservoir computing networks in PDE-based image denoising

  • 期刊论文

  • 发表日期：

    2025年9月4日

查看全文
  • Neda Namaki,
  • Rezvan Salehi,
  • M. R. Eslahchi

摘要

In this work, we improve the results of various partial differential equation (PDE) models, which are used in image denoising, applying reservoir computing networks (RCNs). RCN is a kind of recurrent neural network. Echo state neural network (ESN), which is one type of RCNs, has better quality of denoising and RCNs possess the superiority of shorter training time compared to deep neural networks like convolutional neural network (CNN). The considered PDE models are blending models and consist of isotropic diffusion (ID), Perona-Malik (PM) and some other PDE models, which are applicable in image denoising. In experimental results, the efficiency of using ESN on the output image, which is denoised by blending PDE model, is presented using PSNR criteria comparing with using the PDE model singly.