Smoothing of digital images using the concept of diffusion process

Abstract

An adaptive smoothing algorithm has been described which is capable of performing various tasks, such as removing salt and pepper noise, preserving roof edges, stretching (enhancing) step edges and reducing variations in low intensity varied regions. While iteration advances, it approximates both isotropic and anisotropic heat diffusion processes in performing these tasks. A region topography index has been defined for guiding the algorithm under different situations. Further, an image quality index is proposed which provides a criterion for automatic termination of the algorithm. This criterion can also be used with other iterative smoothing algorithms. The superiority of the method over some other similar techniques has been established for both synthetic and real images.