Fast Imaging Method Based on Curvelet Transform for Radio-interferometric Arrays

Key Points

• The aim is to develop a high-precision imaging method for radio-interferometric arrays that addresses specific challenges in sparse representation and optimization.
• Utilized fast discrete curvelet transform for sparse representation of astronomical images.
• Replaced traditional L1 norm with a smoothed truncated absolute deviation function to approximate L0 norm more accurately.
• Implemented an improved proximal gradient algorithm to solve the non-convex optimization problem.
• Incorporated lazy initialization and gradient-type restart strategies to reduce oscillation and accelerate convergence.
• Conducted multiple simulation experiments and real data validation using specific telescope arrays.
• The proposed method improved reconstruction accuracy compared to existing algorithms.
• Achieved faster computation speed during image reconstruction processes.
• Experimental validations confirmed enhanced performance for future large radio array imaging.