摘要
设计了一种约束的TpV重建模型及其自适应最速下降-投影到凸集求解算法,使用仿真模体和真实CT图像模体进行广泛重建实验,探索在不同的投影稀疏程度下p值的选择对重建精度的影响规律。实验结果表明,p值并非越小越好;p=0.5也不能保证最优;随着投影数据由稀变密,最优p值呈现由1到0逐渐减小的趋势;这些规律可用于对TpV图像重建中p值的选择。
A constrained TpV reconstruction model and its adaptive steepest descent projection onto convex sets algorithm were designed.A wide range of reconstruction experiments were carried out using simulated phantom and the real CT image phantom to explore the influence of p value selection on reconstruction accuracy under different projection sparsity degrees.The experimental results show that the p value is not as small as possible;p=0.5 does not guarantee the optimal;as the projection data changes from sparse to dense,the optimal p value decreases gradually from 1 to 0.These rules can be used to select the p value in TpV image reconstruction.
作者
闫慧文
乔志伟
YAN Hui-wen;QIAO Zhi-wei(School of Computer and Information Technology,Shanxi University,Taiyuan 030006,China)
出处
《核电子学与探测技术》
CAS
北大核心
2021年第6期1039-1044,共6页
Nuclear Electronics & Detection Technology
基金
国家自然科学基金(62071281)
山西省重点研发计划(201803D421012)
山西省留学人员科技活动(RSC1622)
山西省回国留学人员科研(2020-008)资助。
关键词
总变差
最优化
稀疏重建
压缩感知
计算机断层成像技术
Total Variation
Optimization
Sparse Reconstruction
Compressed Sensing
Computed Tomography
