Author(s):
1. Rashid Ali Khan:
Department of Computer Systems Engineering, University of Engineering and Technology,Peshawar,Pakistan
2. Nasru Minallah:
Department of Computer Systems Engineering, University of Engineering and Technology,Peshawar,Pakistan
Abstract:
Diseases like glaucoma, macular degeneration, hypertensive retinopathy and diabetic retinopathy has a major share in complete or partial vision loss in humans. Early diagnose of these diseases is possible through temporal examination of the shape and form, bifurcation patterns and growth of vessels present in the retinal fundus images. In this work, an efficient and rapid scheme to extract the retinal vasculature. In the first step only the green plane is considered and processed out of the RGB color space to accomplish the segmentation process. Contrast enhancement is achieved through applying sigmoid function followed by background exclusion. Finally vessels are extracted through hysteresis thresholding and morphological processing to enhance the fine details in the resultant image. Tradeoff between segmentation accuracy and time consumption of segmentation algorithm is minimized by producing promising accuracy and other metrics. The scheme is tested and evaluated on the retinal fundus images in the databases like STARE and DRIVE. The produced results evaluated and compared to the other state of the art work and proved to be better and outperformed in most cases.
Page(s):
135-144
DOI:
DOI not available
Published:
Journal: Proceedings of the Pakistan Academy of Sciences: A. Physical and Computational Sciences, Volume: 54, Issue: 2, Year: 2017
Keywords:
Diabetic Retinopathy
,
fundus
,
support vector machine SVM
,
Gaussian mixture model GMM
,
sigmoid function
,
contrast limited adaptive histogram equalization CLAHE
,
Markovs random field MRF
References:
References are not available for this document.
Citations
Citations are not available for this document.