Compressed sensing based on L1 and TGV regularization for low-light-level images denoising

Compressed sensing has been applied to image denoising in recent years, and it shows strong noise suppression ability for image corrupted by Gaussian noise. However, the noise contained in low-light-level image is extensive and complex and is modeled as mixed Poisson-Gaussian noise. In this paper, to enable the compressed sensing method to handle such the noise model, the variance-stabilizing transformation and its inverse transformation are used before and after denoising. The total generalized variation constraint term is introduced into the L1 regularization model to maintain the image's structure information, and the alternating direction method of multiplier is used to solve the proposed model. Each subproblem has a closed-form solution. Numerical experiments on artificially degraded and raw low-light-level images show that the proposed method achieves superior performance in terms of visual effects and objective evaluation indices compared with several existing methods.