ICCD low-light-level image denoising based on robust principal component analysis

The intensified charge-coupled device (ICCD) is widely used in the field of low-light-level (LLL) imaging. Compared with ordinary visible light images, ICCD LLL images have lower signal-to-noise ratio (SNR). The ion-feedback noise is a special kind of noise in ICCD LLL images. Because of the noise pollution, it is not good for human eyes to observe and recognize the target details. Therefore, the noise removal of ICCD LLL images is a challenging problem. In this paper, an iterative method based on "detection-location-removal" is proposed. In this method, robust principal component analysis (RPCA) and median filter are used to remove the ion-feedback noise. Before that, Gaussian noise can be suppressed by the block-matching and 3D filtering (BM3D) algorithm. The experimental results show that our method is effective and robust for ICCD LLL image denoising.