An efficient FPGA implementation of Anisotropic Diffusion filter on 2-dimensional and 3-dimensional images

Abstract

Local linear filters like Gaussian filter smooth the whole image irrespective of its edges or fine details, whereas Anisotropic Diffusion is a Non-Linear Filtering approach which considers both image intensity and geometric closeness of the neighboring pixels without smoothing the edges. In this paper, we have worked on the Anisotropic Diffusion filtering and proposed an efficient FPGA design of this filter for both two dimensional and three dimensional images. In this approach, the region boundaries of the image remains sharp and the local operations replicated over to the image making parallel hardware implementations feasible.