Scholarlink

<p>Limited-angle computed tomography (CT) imaging plays a crucial role in reducing radiation exposure and scanning time, particularly in clinical diagnostics. Limited-angle CT images frequently suffer from shading artifacts and structural degradation due to incomplete projection data. The recently developed L1/L2 regularization can preserve image edges and reduce shading artifacts in limited-angle CT reconstruction, it is often computationally intensive. In this work, we use the linearized alternating direction method to solve the L1/L2 regularization problem. The subproblem after linearization can be solved using fast Fourier transform, effectively reducing reconstruction time. The effectiveness of the proposed algorithm is validated through experiments on digital phantoms and real CT data. The results demonstrate that the presented algorithm can suppress shading artifacts and preserve image edges.</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

Limited-angle CT reconstruction by minimizing L1 over L2 based on linearized alternating direction method

  • 期刊论文

  • 发表日期：

    2025年9月16日

查看全文
  • Cheng Qin,
  • Jie Chen,
  • Changcheng Gong

摘要

Limited-angle computed tomography (CT) imaging plays a crucial role in reducing radiation exposure and scanning time, particularly in clinical diagnostics. Limited-angle CT images frequently suffer from shading artifacts and structural degradation due to incomplete projection data. The recently developed L1/L2 regularization can preserve image edges and reduce shading artifacts in limited-angle CT reconstruction, it is often computationally intensive. In this work, we use the linearized alternating direction method to solve the L1/L2 regularization problem. The subproblem after linearization can be solved using fast Fourier transform, effectively reducing reconstruction time. The effectiveness of the proposed algorithm is validated through experiments on digital phantoms and real CT data. The results demonstrate that the presented algorithm can suppress shading artifacts and preserve image edges.