An application of Bottom Hat transformation to extract blood vessel from retinal images

Abstract

Extraction of blood vessels in retinal images provides early diagnosis of different retinopathy diseases (diabetic retinopathy, injury detection, abnormality detection, hemorrhage detection and macular degeneration). This paper presents about the problem of noises and also the blood vessels appearing darker and tiny in the retinal images. This paper introduces a new method for the extraction of retinal blood vessels in retinal fundus images which will be useful to eye specialists in their visual examination of retina and will definitely improve automatic retinal images analysis. In this paper, at first, light reflectance removal technique is used to remove the brighter strips of the images by using green plane of the image. Then, salt and pepper noise and Gaussian noise of the image is removed using median filter and Gaussian filter respectively. After that morphological Bottom Hat Transform is applied to extract the blood vessels. Finally, blood vessels are enhanced using sharpening technique with an unsharp masking. Results are compared with different blood vessel detection algorithms and are found to be encouraging.