Altigen Sayisal Görüntü İşlemede Gri̇-Sevi̇ye Eş-Oluşum Matri̇sleri̇ İle Yüz Tanima

Abstract

Face Recognition has been an attractive field of research for decades, because face is one of the most useful and deterministic biometrics. Image processing is called square pixel-based image processing since its existence. However, hexagonal image processing, which is based on the idea of designing and processing pixels as hexagons, has been shown to provide significant benefits in terms of time and memory savings. Almost all of the face recognition methods proposed and implemented so far are based on square pixel based image processing. Based on the limited number of studies on face recognition in hexagonal pixel based image processing, a hexagonal image processing based face recognition method is proposed in this study. The method proposed in this study is inspired by Grey Level Co-occurrence Matrices (GLCM), which is one of the most fundamental of square pixel based face recognition methods. The method is named Hex_Direct_GLCM because it is based on the square pixel-based basic GLCM method. Since hardwarebased hexagonal pixel-based image processing is not yet available, hexagonal pixel-based equivalents of square pixel-based digital images are artificially created by software. The hexagonal pixel base equivalents of the steps followed in the GLCM method are performed, and then face recognition accuracy performance analysis is performed on different data sets. As presented in the simulation results, the Hex_Direct_GLCM method provides competitive results with high accuracy in terms of face recognition as well as the success in saving resources such as time and space.