Investigation of the Performance of different Spatial Filters toward Mammogram De-noising

Abstract

De-noising is one of the important aspects of image preprocessing, mainly for medical images, in order to filter out the undesired elements without affecting any fine details. In this study spatial filters namely Mean, Median, Wiener and Gaussian filters were employed and the processed images were evaluated for Mean Squared Error (MSE), Peak Signal to Noise Ratio (PSNR) and Correlation Coefficient (CC). The numerical analysis of the accomplished results reveals that Gaussian filter delivered the best outcome for matrix size of 5x5 in terms of all chosen Metrics. Nonetheless, there were not much visual differences among all the filtered images with matrix size 5x5. Although Gaussian filter provided optimal result in this work, however some rooms are left for the improvement in future works using transform domain filters.