The dominance of Poisson noise in color digital cameras

The various components that contribute to the noise in digital cameras are primarily Poisson noise from the discrete nature of the photons, Gaussian and Poisson noise from various electronic and thermal processes and uniform noise from quantization. Our experiments show that the photon Poisson noise is the dominant contributor to uncertainty in the raw data captured by high-performance sensors using in color digital cameras. This finding is consistent with previous monochrome work and indicates that research on denoising and demosaicking of color digital images should consider Poisson noise models as more important than the usual Gaussian models.