APLIKASI PENGENALAN WAJAH PADA MOBILE ROBOT OMNIDIRECTIONAL MENGGUNAKAN METODE PRINCIPAL COMPONENT ANALYSIS (PCA)

Abstract

Self indentification applications have been used in variety cases. Recently in indonesia it was used for e-KTP applications, by taking the characteristics of each person such as fingerprints, signature, retina, and photo of faces. But the applications database have not been optimally used, because those data was just stored offhand.One of data used in e-KTP, photo face could be used as reference in the applications of self indentification. Recognition faces can also reduce cost, as it only requires webcam. The face recognition applications can be used to finding person by combine this applications with mobile robots.

The method used for face recognition are various, there are several criteria that distinguish it by way of the method of identifying face such as: geometric / template based, appearance-based, hybrid appearance based, and models based. Principal component analysis is one of method that was based on appearance criteria. PCA will extract only the important features of data to be used in recognition proses. The product of PCA is eigenface.This eigenfaces from the training process will be stored to database. Euclidean distance will be used to finding the most similar sample with the detected face to recognize it. The mobile robot could detect face and extract the eigenface, then mobile robot would matching it with the database stored in computer to recognize the face.

This method could recognize all the tested face rightly in 80-1300 lux(light intensity metering). If the intensity too dark or too bright the recognition process should fail. Thus the difference of the intensity of training and recognition process couldnt be higher than 1000 lux.

Keyword: e-KTP, principal component analysis, mobile robot omnidirectional.