Scholarlink

<p>Deep image prior, an effective unsupervised deep learning method, has been widely applied to address ill-posed inverse problems, especially in the field of image restoration. Compared to traditional methods, deep image prior achieves superior image recovery results without requiring a large number of labelled samples, but it may lead to some loss of detail. To better preserve the edges and details of images, we propose a novel image denoising model based on deep image prior. This model integrates the <InlineEquation ID="IEq1"> <InlineMediaObject> <ImageObject Color="BlackWhite" FileRef="34_2025_3090_Article_IEq1.gif" Format="GIF" Height="16" Rendition="HTML" Resolution="72" Type="Linedraw" Width="19" /> </InlineMediaObject> <EquationSource Format="TEX">\(L_{1}\)</EquationSource> <EquationSource Format="MATHML"><math> <msub> <mi>L</mi> <mn>1</mn> </msub> </math></EquationSource> </InlineEquation> norm and regularization by denoising into a unified regularizer, fully leveraging the advantages of the deep prior, sparse prior and regularization by denoising prior to better recover image edges and details while effectively mitigating noise. To efficiently solve the model, we utilize a flexible alternating direction method of multipliers to solve the minimization problem. Numerical experiments demonstrate that the new model outperforms others in peak signal-to-noise ratio and structural similarity index, validating its effectiveness.</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

Image Denoising Based on Deep Image Prior Combined Sparsity with Regularization by Denoising

  • 期刊论文

  • 发表日期：

    2025年4月5日

查看全文
  • Jianlou Xu,
  • Yajing Fan,
  • Shaopei You,
  • Li Chen,
  • Yan Hao

摘要

Deep image prior, an effective unsupervised deep learning method, has been widely applied to address ill-posed inverse problems, especially in the field of image restoration. Compared to traditional methods, deep image prior achieves superior image recovery results without requiring a large number of labelled samples, but it may lead to some loss of detail. To better preserve the edges and details of images, we propose a novel image denoising model based on deep image prior. This model integrates the \(L_{1}\) L 1 norm and regularization by denoising into a unified regularizer, fully leveraging the advantages of the deep prior, sparse prior and regularization by denoising prior to better recover image edges and details while effectively mitigating noise. To efficiently solve the model, we utilize a flexible alternating direction method of multipliers to solve the minimization problem. Numerical experiments demonstrate that the new model outperforms others in peak signal-to-noise ratio and structural similarity index, validating its effectiveness.