Nonlinear denoising filter for images with interactive evolutionary computing considering the subjective assessment
Kaoru Arakawa, Kohei Nomoto · 2008
A new type of digital filter for removing random noise in images is proposed using interactive evolutionary computing. This filter is realized as combination of several nonlinear digital filters, such as an epsiv-filter and a conditional median filter. Interactive evolutionary computing is adopted in order to optimize the parameters in the nonlinear filters, considering the subjective assessment by human. Combination of nonlinear filters is usually a powerful tool for noise reduction of images, but such a filter system contains multiple filter parameters which are difficult to be optimized. Interactive evolutionary computing is effective for the total optimization of these filter parameters. Moreover, human taste and subjective sense are highly considered in the filter performance. As an example of this type of filter, a cascade of two epsiv-filters and a conditional median filter is presented for reducing random noise, and its computer simulations are shown to verify its high performance.