Optimal Estimation of the Position of Non-Gaussian Impulse Noise in Images
E. A. Samoilin
Keywords: pulse noise, optimal estimation, spatial position, image filtering, average risk
Pages: 74-82
Abstract
An algorithm for optimal estimation (in the sense of the minimum average risk criterion) of the spatial position of non-Gaussian impulse noise in digital images is proposed based on the signal difference between the center of the slide aperture and its other elements. An expression for the boundary values of the admissible signal interval is obtained that provides the potentially achievable zero probability of noise estimation error. Results of numerical studies of the efficiency of the synthesized algorithm show its advantages over the well-known nonoptimal algorithm.
|