A Comparative Study on Medical Image Denoising in Hybrid domain

The key to medical image denoising technique is to remove the noise while preserving important features. Non-local mean filtering and bilateral filtering are commonly used procedures for medical image denoising. In this paper analysis and comparison of spatial as well as frequency domain methods including bilateral filtering , non-local mean filtering, wavelet thresholding, contourlet thresholding and non-subsampled contourlet transform are done. The non-subsampled contourlet transform applied prior to bilateral and non-local mean filtering gives improved PSNR and perceptual quality. From the results it is also found that introducing transform domain method prior to spatial domain method does not increase the processing time to a much extend.