Optimasi Age Invariant Face Recognition dengan Image Pyramid Menggunakan OD-LBP dan LBP

Abstract

This research was motivated by problems with the facial recognition system which had difficulty recognizing individuals of various ages. This is because human age will often increase and faces will increasingly change as time goes by. Therefore, Age Invariant Face Recognition (AIFR) is very important to develop in the field of technology to identify missing people or criminals. The aim of this research is to optimize AIFR by using various image extractions and using image pyramids and to find out which extraction method is more effective. This research uses a replication of previous research, namely Local Binary Pattern (LBP) and as a comparison is Orthogonal Difference-Local Binary Pattern (OD-LBP). There are 2 datasets for AIFR, namely FGNET and MORPH, which were the material of previous research. However, this research will use the FGNET dataset as a consideration with previous research. Utilization of Support Vector Machine (SVM) will be used as a classification method. The results of this study show the highest accuracy in the combination of the OD-LBP method with all facial image extraction as the image extraction. However, LBP with facial segments achieves low accuracy with short extraction time. This research provides sufficient contributions to the selection of suitable methods according to the needs of AIFR.