Comparison of Pre-Processing Methods for Segmentation and Approximation of Optic Disc Boundary from Processed Digital Retinal Images

Comparison of Pre-Processing Methods for Segmentation and Approximation of Optic Disc Boundary from Processed Digital Retinal Images

Algorithms are presented for rapid, automatic and accurate segmentation and Approximation of Optic Disc boundary from digital retinal images. This paper presents method that improves upon prior work in different ways; 1.Accuracy 2.Contrast. The pre-processing methods are presented here to enhance th...

Full description

Saved in:
Bibliographic Details
Published in:2014 International Conference on Devices, Circuits and Communications (ICDCCom) pp. 1 - 6
Main Authors: Vaidya, Yogita M., Doiphode, Sunil E.
Format: Conference Proceeding
Language:English
Published: IEEE 01-09-2014
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Algorithms are presented for rapid, automatic and accurate segmentation and Approximation of Optic Disc boundary from digital retinal images. This paper presents method that improves upon prior work in different ways; 1.Accuracy 2.Contrast. The pre-processing methods are presented here to enhance the image properties prior to the segmentation of OD from digital fundus images of retina using template based methodology. This paper compares performance of contrast enhancement techniques such as Intensity thresholding, adaptive histogram equalization and histogram equalization. Further, evaluates performance of each of these techniques with respect to the original template based OD segmentation using circular Hough Transform. Intensity thresholding method provides better performance with Maximum Variance method and Low Pass Filter method of ODP location. Histogram Equalization performs better with only Low Pass Filter method of ODP location. Adaptive Histogram Equalization performs better with all three methods of OD pixel location so ultimately it performs better with voting type algorithm that is used to locate the centroid of the pixel location detected by Maximum Difference; Maximum Variance and Low Pass filter method. Thus the optic disc segmentation performance is improved over the original method. Experimental results on a known MESSIDOR database and local NIOP database, achieving to more than 93% accuracy for optic disc boundary approximation. it proved that adaptive histogram performs better as a contrast enhancement technique.
DOI:10.1109/ICDCCom.2014.7024709