Taner Cevik

Work place: Computer Engineering, Istanbul Aydin University, Istanbul, Turkey

E-mail: tanercevik@aydin.edu.tr


Research Interests: Multimedia Information System, Image Processing, Computer Networks, Computer systems and computational processes


Taner Cevik received the B.Sc. degree in computer engineering from Istanbul Technical University, Istanbul in 2001, and obtained his Master of Science degree in computer engineering from Fatih University, Istanbul in 2008. Following, completed Ph.D. at Istanbul University in 2012. He joined the Computer Engineering Department at Aydin University in 2017 and continues to work as an Associate Professor. His research interests are Image Processing and Wireless Multimedia Sensor Networks .

Author Articles
Discrete Wavelet Transform based High Performance Face Recognition Using a Novel Statistical Approach

By Nazife Cevik Taner Cevik

DOI: https://doi.org/10.5815/ijigsp.2018.06.01, Pub. Date: 8 Jun. 2018

Biometrics has gained significant popularity for individual identification in the last decades as a necessity of supporting especially the law enforcement and personal authentication required applications. The face is one of the distinctive biometrics that can be used to identify an individual. Henceforth, Face Recognition (FR) has attracted the great interest of the scientists and academicians. One of the most popular methods preferred for FR is extracting textual features from face images and subsequently performing classification according to these features. A substantial portion of the previous texture analysis and classification studies have based on extracting features from Gray Level Co-occurrence Matrix (GLCM). In this study, we present an alternative method that utilizes Gray Level Total Displacement Matrix (GLTDM) which holds statistical information about the Discrete Wavelet Transform (DWT) of the original face image. The approximation and three detail sub-bands of the image are first calculated. GLTDMs that are specific to these four matrices are subsequently generated. The Haralick features are extracted from those generated four GLTDMs. At the following stage, a new joint feature vector is formed using these four groups of Haralick features. Lastly, extracted features are classified by using K-NN algorithm. As demonstrated in the simulation results, the proposed approach performs promising results in the context of classification.

[...] Read more.
Other Articles