Automatic segmentation of the blood vessels from the fundus image and analysis of the retinal vascular tree are useful in computer aided diagnosis. Any change in the branching pattern, dimension and tortuosity of the blood vessels indicates the degree of severity of disease in the body and the eye. In this paper the retinal color image is decomposed into its RGB channels and a 2D matched filter is used for enhancing the blood vessels. Using an unsupervised classification technique,the blood vessels are segmented.The membership function in spatial FCM includes the spatial neighborhood information that is used for clustering Finally, these outputs are fused together using the maximization technique and tested over the DRIVE database. The results are comparable with other methods and provide a reliable methodology for vessel pattern analysis. © 2015 Research India Publications.