Face recognition system based on local binary pattern and gray level co-occurrence matrix

Biometrics is defining people according to their physiological, behavioral and biological characteristics. Biometrics can be divided into two categories: physiological biometric and behavioral biometric. Physiological biometrics are features such as face, iris, fingerprint finger vessels, hand geometry, etc. that help identify the individual with their physiological or biological features. Behavioral biometrics, on the other hand, are features such as handwriting, signature, or tone of voice that gain individual dimensions over time and thus help to recognize the individual. In this thesis, a discrete wavelet transform is presented based on local binary patterns, gray level and co-sequencing matrix. It represents a new approach to face datasets. The proposed face recognition system has been designed by considering different usage purposes. Wavelet transform was used to compress face training data, then extraction of features from face images using local binary patterns based on collocation matrix. Performance evaluation of the proposed method has been made on ORL databases. In the thesis study, it has been shown that face recognition systems have a great effect on the improvement of the results in the preprocessing stage.