Retinal identification system based on properties of image texture

among the methods of biometric, retinal images are the most accurate biometric therefore retinal identification system is employed in environments, where extremely highsecurity measures are needed. A new biometric identification based on texture properties of images presented in this paper.The propose system is supposed to extract feature by the use of combining statistical and spectral approaches of texture. Cooccurrencematrices and their properties are computed for feature extraction via statistical approaches. Afterward Fourier transform is applied to achieve the Fourier spectrum of retinaimage and feature vectors are made based on the description of spectrum with two functions in polar coordinate. The proposedmethod is tested on a database consist of 280 images obtained from 40 subjects of DRIVE database. Experimental results are reported in last section.

کلیدواژه ها:

Identification ، retinal image ، Fourier Spectrum ، co-occurrence matrix