Title: New circularity index using Radon transform
Authors: Mohammed Ammar; Saïd Mahmoudi
Biomedical Engineering Laboratory, University of Tlemcen, Algeria; Physics Department, Faculty of Sciences, University of Boumerdes, Algeria
Computer Science Department, Faculty of Engineering, University of Mons, 20 Place du Parc, Mons 7000, Belgium
Abstract: In this paper, we propose a new shape descriptor in order to evaluate the circularity index on the basis of the Radon transform. In this study, we focus on the creation of a Radon-based circularity index. This index takes its values between 0 and 1, and picks the value of 1 if and only if the measured shape is a circle. This new circularity index performs better for shapes with boundary defects which lead to a large increase in perimeter, and also in the case of hollow circular shapes. We compare the proposed index with the classical circularity measures and Hu moment invariants based method by using some structures of interest in medical images like cell nucleus in cytological images, skin tumours and cardiac MRI images. The new measure is robust to different kinds of noise: Gaussian, salt, pepper and speckle noise.
Keywords: shape descriptors; Radon transform; circularity index; medical shapes; medical images; cell nucleus; cytological images; skin tumours; cardiac images; MRI scans; magnetic resonance imaging; magnetic resonance images.
Int. J. of Applied Pattern Recognition, 2016 Vol.3, No.4, pp.368 - 380
Submission date: 27 Apr 2016
Date of acceptance: 12 Aug 2016
Available online: 09 Feb 2017