Blood Vessel Segmentation of Fundus Images by Major Vessel Extraction and Subimage Classification - 2015
This paper presents a unique 3-stage blood vessel segmentation algorithm using fundus pictures. In the primary stage, the inexperienced plane of a fundus image is preprocessed to extract a binary image when high-pass filtering, and another binary image from the morphologically reconstructed enhanced image for the vessel regions. Next, the regions common to each the binary images are extracted as the major vessels. Within the second stage, all remaining pixels in the two binary pictures are classified employing a Gaussian mixture model (GMM) classifier using a set of eight options that are extracted based mostly on pixel neighborhood and first and second-order gradient pictures. In the third postprocessing stage, the foremost portions of the blood vessels are combined with the classified vessel pixels. The proposed algorithm is a smaller amount dependent on training data, needs less segmentation time and achieves consistent vessel segmentation accuracy on traditional pictures in addition to images with pathology compared to existing supervised segmentation methods. The proposed algorithm achieves a vessel segmentation accuracy of 95.two%, ninety five.15percent, and 95.three% in an average of three.one, six.7, and eleven.7 s on 3 public datasets DRIVE, STARE, and CHASE_DB1, respectively.
Did you like this research project?
To get this research project Guidelines, Training and Code... Click Here