Title: Iterative modelling of the closing-based differential morphological profile

Authors: Arif Muntasa; Indah Agustien Siradjuddin

Addresses: Computational Artificial Intelligence Laboratory, Informatics Department, University of Trunojoyo Madura, Ry, Telang Po. Box 2, Kamal, Bangkalan, East Java, Indonesia ' Computational Artificial Intelligence Laboratory, Informatics Department, University of Trunojoyo Madura, Ry, Telang Po. Box 2, Kamal, Bangkalan, East Java, Indonesia

Abstract: One of an image processing applications is retinal image optic disk detection. The similarity of the retinal gray scale image between the object and background has been interesting to many researchers to develop the research. In this research, the closing-based differential morphological profile is proposed to detect the optic disc on the retinal image. The closing process is performed by iterative. It is started by pre-processing and followed by the different morphological profile based on the closing operation, i.e., the dilation and erosion processes. The dilation process is iteratively performed and followed by erosion process. The process results are enhanced to obtain the better image when the binary image transformation is conducted. The noise removal process is also necessity to eliminate the detection error. Furthermore, determining of the point centre of an object detection result will be used to create the optic disk. The detection rates of the proposed approach show that the maximum detection accuracy outperformed to the other methods, i.e., 2D-Gaussian filtering-based mathematical morphology approach, differential morphological profile, morphological reconstruction techniques and hybrid fuzzy classifier.

Keywords: differential morphological profile; iterative modelling; optic disk image detection; closing operation.

DOI: 10.1504/IJBET.2019.101053

International Journal of Biomedical Engineering and Technology, 2019 Vol.31 No.1, pp.84 - 103

Received: 14 Sep 2016
Accepted: 01 Feb 2017

Published online: 23 Jul 2019 *

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