Nonlinear filters based parallel demosaicking and denoising of natural images infected with non Gaussian noise
Barre Prabhakar Rao, Shraddha Prasad · 2022
In the process of image restoration from images corrupted with non-Gaussian noise, the nonlinear filters perform superior over linear filters, and over the conventional sequential demosaicking and denoising methods, the parallel demosaicking cum denoising is highly efficient and fast. These assumptions in this paper are analysed, implemented, evaluated and established objectively by the authors. Most the research is found to be non pragmatic as the restoration of the images from non-Gaussian noise infected images is rarely considered. Hence the authors, by implementing sixteen nonlinear filter techniques and one linear filter technique, reconstructed the color from the images infected with Gaussian and Non- Gaussian Noise.Objective performance metrics MMSE, CPSNR, and SCIELAB are obtained and compared for objective performance evaluation, and the reconstructed images are also presented for the viewers’ subjective evaluation. The standard image dataset is used for implementation of the proposed nonlinear filter based parallel demosaicking cum denoising of non-Gaussian noise infected images.