Images corrupted by gaussian noise enhancement via piecewise linear models

摘要

Noise having Gaussian-like distribution is very often encountered during image acquisition. Tranditional enhancement methods (space domain and frequency domain) can reduce noise, but also are easy to produce blurred edge. So 5-parameter PWL nonlinear transformation is used to enhance image and reduce noise. A satisfactory combination of image enhancement and noise attenuation can be achieved by different parameter settings. Results dealing with images with different noise variance show that the method yields a very effective enhancement of the image data.