“…With the rapid development of complementary metal oxide semiconductor (CMOS) technologies, CMOS image sensors have become popular with consumer and vehicle electronics, telemedicine, video surveillance, space exploration, fluorescence detection, and so on [1][2][3][4]. However, images generated by these sensors inevitably contain noise, owing to their internal structure, which results in image quality degradation [5,6], and thus, estimating these noise parameters accurately assumes paramount importance in improving the performance of denoising algorithms [3,4,[7][8][9]. For CMOS image sensors, a signaldependent noise model, such as the Poisson-Gaussian model, can more accurately delineate the noise characteristics than an additive channel-dependent noise model [10][11][12][13][14][15][16][17][18][19].…”