Image Noise Removal for Embedded Devices - Characterization, estimation and filtering techniques

This thesis aims to provide an overview of the complex problem of noise in digital images, yielding an excursus on the principal noise sources and analyzing some classical noise estimation and filtering algorithms. Useful and innovative methods for the characterization of the imager signal dependent noise are then introduced, jointly with some reliable and effective signal dependent noise reduction algorithms. In addition, a novel denoising technique based on the study of the human visual system is presented.