An Exploration on Various Nonlinear Filters to Preserve the Edges of a Digital Image in Spatial Domain
Seema S. Bhat, M. C. Hanumantharaju, M. T. Gopalakrishna · 2015
Preservation of details in an image while denoising is a significant part in the preprocessing stage of digital image processing. Since the past two decades, various filters have been proposed by numerous researchers to address the common problem of preserving the edges during denoising. Most of the edge preserving denoising filters are suitable only for low contrast images and it's applicability has progressed significantly for variety of images. This survey provides a compact summary of the different nonlinear filters for details preservation in the spatial domain. Some of the filters decline from intrinsic theoretical limitations that triggers dramatic instability in the presence of impulsive noise and shadowed images. The comparisons of visual results of the filters demonstrates similar outputs, but often deteriorate with artifacts and halos. As is evident from the review of related work, the Guided image filter is agile and extensively used edge preserving denoising filter. They are beneficial in edge preserving denoising, high dynamic range compression, image feathering, haze removal, joint upsampling and many such applications as their computational complexity is independent of the filtering kernel size.