Nonon-Convex Low-Rank Minimization for Sparse-View CT Reconstruction via Nonlocal-Group Dictionary Learning
2020
Recently, low-rank minimization has demonstrated its success in sparse-view CT reconstruction. However, traditional nuclear norm minimization (NNM) methods often treat all singular values equivalently and lead to biased solution inevitably. In this paper, we propose a nonconvex low-rank minimization approach for SART based Sparse-View CT reconstruction via group dictionary learning scheme. To further improve CT image quality, we develop a nonlocal-group based low-rank minimization for CT image enhancement via ADMM framework. Two effective non-convex surrogates of weighted schatten p-norm and weighted truncated nuclear norm are utilized for more accurate approximation of the low-rank group. Experimental results demonstrate that our method shows a better performance than previous work.
Keywords:
- Correction
- Source
- Cite
- Save
- Machine Reading By IdeaReader
0
References
0
Citations
NaN
KQI