Abstract

Aalgorithms for facial recognition PCA (Principal component analysis) encounter a dramatic performance decline when analyzing photos taken from various angles. As part of this research, a 3-D model was utilized for the sake of improving abovementioned performance. This model enables us to convert an image of somebody2019;s face, which was taken from a specific angle to another face. Model was tested using biometric database created on Belgrade University. Picture rotation according to model was performed before comparing with the specimen pictures from the database. Results of this study have shown accuracy in regards to biometric verification, where identification percentage has been drastically improved.