Small Retinal Blood Vessel Tracking Using an Adaptive Filter

Abstract

The problem of extracting small retinal blood vessels using a Canny edge detection method is addressed. It shows that using the Canny edge detector to detect retinal blood vessels, especially small vessels, can be significantly improved if an adaptive edge detection filter is incorporated. The filter is designed as a local dynamic hysteresis thresholding value generator. It adapts knowledge of the location of major vessels to define a small neighborhood and to generate the local hysteresis threshold values to detect meaningful edges, especially the edges of small blood vessels that may be missing, using Canny edge detector alone. The effectiveness of the adaptive edge detection filter is demonstrated by the preliminary experimental results obtained with the proposed method. A comparative test is also presented to highlight the performance differences between the Canny edge detector with the adaptive edge detection filter and the one without the filter.