Gaussian Noise Estimation in Digital Images Using Nonlinear Sharpening and Genetic Optimization
F. Russo · Conference proceedings - IEEE Instrumentation/Measurement Technology Conference · 2007
A new approach to estimation of Gaussian noise in digital images is presented. As a first step, a nonlinear amplification of the noise is provided by adopting a multiparameter piecewise linear (PWL) sharpener. Thus, the noise is estimated by analyzing the edge gradients of the data filtered by a PWL smoother. The optimal parameter values for the sharpening stage are found by resorting to a simple genetic algorithm and a set of training data. Computer simulations show that the approach gives accurate results in a wide range of noise variances.