Abstract

GARCIA GARCIA, Yainet; RODRIGUEZ GUILLEN, Reinier  and  TABOADA CRISPI, Alberto. Mapping of digital images of eye fund attending to features of texture. Rev cuba cienc informat [online]. 2017, vol.11, n.1, pp.106-121. ISSN 2227-1899.

ABSTRACT The present work has the interest to contribute in the automatic analysis of digital images of eye background. The aim of this study is to define an optimal selection of texture features that facilitate the location of the optic disc and the segmentation of blood vessels. The MaZda software allows the mapping of almost 300 texture features by combining 6 statistical descriptors: histogram, gradient, Run-Length matrix, co-occurrence matrix, Wavelet transform and autoregressive model. The mapping of 150 images belonging to the Databases DRIVE and DIARETDB0 is carried out. The results obtained are encouraging because they could allow blood vessel segmentation to be 100% successful and 99.3% successful for the location of the optic disc. Thus we obtain an efficient selection of traits for the development or improvement of algorithms with a minimum pre-processing.

Keywords : Optical disc; eye background; mapping of digital images; MaZda; texture features; blood vessels.

        · abstract in Spanish     · text in Spanish     · Spanish ( pdf )