DefinePK

Abstract

The de-noising of digital images is crucial preprocessing step before moving toward image segmentation, representation and object recognition. It is important to find out efficacy of filter for different noise models because filtering operation is application oriented task and performance varies according to type of noise present in images. A comparative study has made to elucidate the behavior of different spatial filtering techniques under different noise models. In this paper different types of noises like Gaussian noise, Speckle noise, Salt & Pepper noise are applied on grayscale standard image of Lenna and using spatial filtering techniques the values of full reference based image quality metrics are found and compared in tabular and graphical form. The outcome of comparative study shows that Lee, Kuan and Anisotropic Diffusion Filter worked well for Speckle noise, the Salt and Pepper noise has significantly reduced using Median and AWMF, and the Mean filter and Wiener filter works immensely efficient for reducing Gaussian noise.