Performance Analysis of Various Spatial Filters on Noisy Images - A Quantitative Approach

摘要

Choosing a suitable denoising filter acts a significant role in the image restoration procedure. Smoothing or reduction of noise present in the image is a significant preprocessing task before segmenting and analyzing the features present in an image. This paper is focused on choosing a desired filter for performing denoising process through some of the well-known image quantitative metrics. Some of the traditional filters are taken for experimentation and quantitative results are compared. In order to evaluate the degree of smoothness of the resultant images, quantitative metrics such as MSSIM (Mean Structural Similarity Index), PSNR (Peak Signal-to-Noise Ratio), and MSE (Mean Square Error) are used. This paper gives guidance in selecting an appropriate filter for performing denoising process through quantitative metrics.