Face Recognition using Two Local Ternary Patterns (LTP) Variants: A Performance Analysis using High-and Low-Resolution Images

Abstract

Face recognition has grown over the years and has become a popular field of research in computer vision. It has gained prominence due to many successful applications in security, surveillance, healthcare, and other areas. Many face recognition techniques that focuses on facial texture features have also been widely proposed. Amongst the most commonly used techniques for face recognition based on image texture features are the Local Binary Patterns (LBP), Local Ternary Pattern (LTP) and their variations. Face texture features are more influenced by image resolution. Recognition accuracy tends to degrade with changes to image resolution. Therefore, image resolution is an important factor for accurate face recognition. In this paper, two face texture feature based techniques (DLTP & ELTP) are analyzed to determine how they perform when image resolution changes. This study provides a comprehensive analysis of experimental results for two Local Ternary Patterns (LTP) on how they are affected by image resolution changes. Results suggest that low and high resolution versions of the same image do not hold the same information and hence when compared the two are likely to be classified differently.