This work takes advantage of multispectral data from human irises, ocular prostheses and colored contact lenses to compare them accounting for their texture features. For this aim, several procedures from image processing were implemented: image segmentation algorithms to extract the areas of interest, statical texture analysis tools for texture description, and statistical classifiers for samples discrimination from texture descriptors. For the texture study two approaches were implemented, firstorder statistical study of the image histogram and second order statistical analysis through co-occurrence matrices. The results of the samples classification regarding the texture show differences in color distribution among samples. Thus, it can be stated that the comparison of just color features of the samples are not enough to conclude whether a contact lens or ocular prosthesis is a good reproduction of an iris sample, and therefore, it is necessary to include the texture in such comparisons.

Peer Reviewed