Performance improvement of decision median filter for suppression of salt and pepper noise

Abstract

Integration of decision based schemes with median filtering has been applied previously in numerous works to identify and process only the corrupted pixels during image denoising. However, these approaches are performance limited owing to their dependency upon selection of pre-defined thresholds as a decision measure. This paper presents a novel algorithm for performance improvement of decision median filter for suppression of salt and pepper noise in digital images. The proposed algorithm performs decision (to adaptively increase the window size) by comparing the computed median with the minimum and maximum pixel values within a local window. Thereafter, the pixels are processed with the proposed algorithm; reaching a maximum window size limit of 9x9. Objective analysis of the obtained results is carried out using Peak Signal to Noise Ratio (PSNR) and Structural Similarity Index (SSIM) as quality parameters. As depicted from the simulations results the proposed algorithm is capable to suppress noise effectively even with the noise contamination levels as high as 90%. © Springer International Publishing Switzerland 2014.