Research into CUP-VISAR velocity reconstruction based on weighted DRUNet and total variation joint optimization
Xi Wang, Lei Zhang, Miao Li, Boshan Yu, Zhaohui Guo, Xueyin Zhao, Feng Wang, Yulong Li, Zanyang Guan
Abstract
This Letter proposes a CUP-VISAR data reconstruction algorithm for laser-driven inertial confinement fusion (ICF) research. The algorithm combines weighted deep residual U-Net (DRUNet) and joint optimization with total variation (TV) to improve shockwave velocity fringe image reconstruction. The simulation results demonstrate that the proposed algorithm outperforms the ADMM-TV and enhanced 3D total variation (E-3DTV) algorithms, enhancing the quality of the reconstructed images and thereby improving the accuracy of velocity field calculations. Furthermore, it addresses the challenges of the high compression ratio caused by the diagnostic requirements of the larger number of sampling frames in the CUP-VISAR system and the issues of aliasing within a large encoding aperture. The proposed algorithm demonstrates good robustness to noise, ensuring reliable reconstruction even under Gaussian noise with a relative intensity of 0.05. This algorithm contributes to ICF diagnostics in complex environmental conditions and has theoretical significance and practical application value for achieving controlled thermonuclear fusion.