Cyclostationary Gaussian noise: theory and synthesis

Abstract Stationary Gaussian processes have been used for decades in the context of procedural noises to model and synthesize textures with no spatial organization. In this paper we investigate cyclostationary Gaussian processes, whose statistics are repeated periodically. It enables the modeling of noises having periodic spatial variations, which we call “cyclostationary Gaussian noises”. We adapt to the cyclostationary context several stationary noises along with their synthesis algorithms: spot noise, Gabor noise, local random‐phase noise, high‐performance noise, and phasor noise. We exhibit real‐time synthesis of a variety of visual patterns having periodic spatial variations.