Most denoising methods assume additive white Gaussian noise. As camera noise is fundamentally different from this common noise model, this dissertation proposes denoising algorithms adapted to real camera data. A new parameter estimation method is proposed to achieve image-adaptive parameter tuned for high visual quality. For improving the evaluation of denoising methods with respect to real camera noise, this thesis proposes a realistic evaluation method and analyzes the human perception of noise in video data.
«
Most denoising methods assume additive white Gaussian noise. As camera noise is fundamentally different from this common noise model, this dissertation proposes denoising algorithms adapted to real camera data. A new parameter estimation method is proposed to achieve image-adaptive parameter tuned for high visual quality. For improving the evaluation of denoising methods with respect to real camera noise, this thesis proposes a realistic evaluation method and analyzes the human perception of noi...
»