Segmentation Of Vessel Structure In Retinal Images By Matched Filtering

A method for automatic segmentation of the vessel structure in colour images of retina provided by digital fundus camera is presented. The method utilises 2D matched filtering to detect presence of short linear vessel sections of a particular thickness and direction; the approach correlates the respective (small) image areas with a 2D mask based on a typical brightness profile perpendicular to vessels of a particular width. Three different approximated profiles are used, e.g. for thin, medium and thick vessels; the corresponding masks are provided by 2D extension and rotation into 12 orientations. The parametric images obtained this way are then thresholded in order to obtain binary images depicting the presence of vessels of a particular width. The three binary representations are consequently combined to provide the best available rough vessel map, which is finalised by complementing the obviously missing vessel sections and cleaning the disconnected fractional artefacts. The method has been developed using our database of real retinal images and finally tested on images from the international DRIVE database of retinal images for which standard vessel maps derived medical experts are available.