Efficient Denoising Framework for Mammogram Images with a New Impulse Detector and Non-Local Means

OBJECTIVE: The survival rates of breast cancer are increasing as screening and diagnosis improve. The removal of noise is revealed to be a significant step for automatic - computer aided detection (CAD) of microcalcification in digital mammography. METHODS: In this paper, a combined approach for eradicating impulse noise from digital mammograms is proposed. The process is achieved in two stages, detection of noise followed by filtering of noise. The detection of noise is carried out by using Modified Robust Outlyingness Ratio (mROR) trailed by an extended NL (Non-Local)-means filter for filtering mechanism. RESULTS: According to the value of mROR, all pixels in mammogram images are divided into four distinct groups. In each cluster, many decision rules are then applied for detecting the impulse noise. Filtering is done with NL-means filter by providing a reference mammogram image. CONCLUSION: The comparative analysis and evaluated results are compared with some existing filters which indicate that the proposed structure outperforms the analysed result of others.