Abstract: Blood vessel retinal image segmentation has important application value in clinical medical. According to the method of retinal blood vessels segmentation, a new and effective method of retinal vessel segmentation is proposed based on MSLTA(Multi-Scale Single-Channel Linear Track) image segmentation. Pretreatment by Gabor filtering enhances blood vessel information, and obtains initial vascular network algorithm with MSLTA. The method of connected component labeling is used to remove noise in order to remove the speckle noise of the image using the Retinal image of the international public DRIVE (Digital Retinal Images for Vessel Extraction) database to experiment,and compared with conventional existing algorithms, the experimental results showed that the average precision of this method reaches 95. 37% and the micro vessel is also able to be preserved.

Key words: retinal image, vessel segmentation, multi鄄scale single鄄channel linear track, connected component labeling