Turkish Journal of Electrical Engineering and Computer Sciences
A retinal image contains vital information. Extracting these features is the first and most important step in the analysis of retinal images for various applications of medical or human recognition. In this study, a morphological-based blood vessel extraction algorithm using adaptive-local analysis from colored retinal images is proposed. In this algorithm, by applying the appropriate morphology functors and local histogram stretching on the retinal images, the brightness of the images is considerably uniformed. Furthermore, curvelet transform (CT) is used to enhance the retinal images by highlighting the edges of the images in various scales and directions and by the adaptive and local improving of the CT coefficients. Since the blood vessels in retinal images are distributed in various directions, we use the morphology functors with multidirectional structure elements to extract the blood vessels from the retinal images. Geodesic conversion-based morphology functors are used to properly refine the appeared frills with a size smaller than those of the arterioles in the images. Finally, by locally applying the connected component analysis in the images and locally applying an adaptive filter on the connected components, all of the residual frills are refined from the images. The obtained results of the proposed algorithm show that the blood vessels are extracted from the background of the images with high accuracy, which in turn shows the high ability of the proposed algorithm in extracting the retinal blood vessels.
Blood vessel extraction, retinal image, curvelet transform, adaptive-local analysis, geodesic conversion-based morphology functors, multidirectional morphology functors
"Retinal image analysis using multidirectional functors based on geodesic conversions,"
Turkish Journal of Electrical Engineering and Computer Sciences: Vol. 22:
3, Article 19.
Available at: https://journals.tubitak.gov.tr/elektrik/vol22/iss3/19
Computer Engineering Commons, Computer Sciences Commons, Electrical and Computer Engineering Commons