Title: A new and efficient approach for the removal of high density impulse noise in mammogram

Authors: S. Sreedevi; Elizabeth Sherly

Addresses: Department of Computer Science, Sree Ayyappa College, Eramallikkara P.O., Chengannur, Alleppey(dt), Kerala, 690-574, India ' Indian Institute of Information Technology and Management – Kerala, Technopark, Thiruvanandapuram, Kerala, 695-581, India

Abstract: This paper proposes a combined approach for removing impulse noise from digital mammograms which implement a detection followed by filtering mechanism, in which, detection is done using a robust local image statistical measure called modified robust outlyingness ratio (MROR) followed by a filtering framework based on extended nonlocal means (ENLM). All the pixels in the image are grouped into four different clusters based on the value of MROR. The detection system consists of two stages, coarse stage and fine stage. In each stage, different decision rules are adopted to detect the impulse noise in each cluster and to restore the image, the value of the noisy pixels is replaced with the modified median-based value of the corresponding window based on the cluster position. For filtering, the NL-means filter is extended by introducing a reference image. Simulations are carried out on the MIAS database and the performance of the proposed filter has been evaluated quantitatively and qualitatively through experimental analysis and the results are compared with several existing filters such as standard median filter (SMF), adaptive median filter (AMF), robust outlyingness ratio – non local means (ROR-NLM) and modified robust outlyingness ratio – non local means (MROR-NLM).

Keywords: impulse noise; image denoising; non-local means filter; noise detector; ROR; adaptive median filter; AMF; coarse stage; fine stage; MROR-ENLM.

DOI: 10.1504/IJCAET.2020.106247

International Journal of Computer Aided Engineering and Technology, 2020 Vol.12 No.3, pp.370 - 391

Received: 10 Oct 2017
Accepted: 01 May 2018

Published online: 05 Mar 2020 *

Full-text access for editors Access for subscribers Purchase this article Comment on this article