PATTERN RECOGNITION FROM FACE IMAGES

Authors : Tolga Ensari

Pages : 14-20

View : 13 | Download : 10

Publication Date : 2017-11-30

Article Type : Research

Abstract :In this article, we use projected gradient descent nonnegative matrix factorization (NMF-PGD) method and make pattern recognition analysis on ORL face data set. Face recognition is one of the critical issues in our life and some security, daily activities and operations use this well known application area. NMF-PGD is a type of nonnegative matrix factorization (NMF) which defined in the literature. In the study, derived NMF-PGD definition and algorithm has been used in order to classify the ORL face images. We give the experimental results in a table and graph. According to experiments, face recognition accuracy rates have different accuracy values because of the k - lower rank value. We change k-values between 25 and 144 to see the performance of NMF-PGD. At the end, we make some analysis and comments on the recognition rates. Additionally, NMF-PGD can also be used for different kind of pattern recognition problems.
Keywords : Pattern Recognition, Classification, Face Recognition, Nonnegative Matrix Factorization