Scholarlink

<p>In response to existing compressed sensing ghost imaging (CSGI) schemes, an innovative Bayesian compressed sensing ghost imaging with better anti-noise performance is proposed, by using the sparse representation of K-Singular Value Decomposition (KSVD) and a 3Level (3L)-hierarchical variational message passing (VMP) algorithm. Simulation and experimental results confirm that, this innovative method overcomes the limitations of presetting specific parameters (sparsity, noise level, etc.), and also demonstrates superior performance in terms of reconstruction accuracy and imaging quality, especially for highly complex objects, where it effectively achieves accurate imaging under varying levels of noise at a low sampling rate (below 12.2%). In addition, compared to existing Bayesian compressive sensing ghost imaging (BCSGI), our algorithm moderately reduces time consumption while ensuring high precision. Our results may provide potential applications of CSGI in the field of biomedical imaging.</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

Anti-noise variational sparse Bayesian estimation ghost imaging based on 3Level factor graph

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年12月29日

查看全文
  • Siqing Xiang,
  • Yanfeng Bai,
  • Qi Zhou,
  • Jianxia Chen,
  • Jintao Zhai,
  • Jiaqi Yin,
  • Tengfei Liu,
  • Xuan Liu,
  • Xianwei Huang,
  • Xiquan Fu

摘要

In response to existing compressed sensing ghost imaging (CSGI) schemes, an innovative Bayesian compressed sensing ghost imaging with better anti-noise performance is proposed, by using the sparse representation of K-Singular Value Decomposition (KSVD) and a 3Level (3L)-hierarchical variational message passing (VMP) algorithm. Simulation and experimental results confirm that, this innovative method overcomes the limitations of presetting specific parameters (sparsity, noise level, etc.), and also demonstrates superior performance in terms of reconstruction accuracy and imaging quality, especially for highly complex objects, where it effectively achieves accurate imaging under varying levels of noise at a low sampling rate (below 12.2%). In addition, compared to existing Bayesian compressive sensing ghost imaging (BCSGI), our algorithm moderately reduces time consumption while ensuring high precision. Our results may provide potential applications of CSGI in the field of biomedical imaging.