This paper presents a new technique, named Improved Local Line Binary Pattern (ILLBP), which is an operator for illumination-robust face recognition from a single training image. In order to empirically demonstrate effectiveness of the proposed approach, we use Principal Component Analysis-Nearest Neighbour (PCA-NN) and multi-class Support Vector Machine (SVM) as the classifiers. Comparisons to the Local Line Binary Pattern (LLBP) on Yale Face Database B are also conducted. The advantages of our technique include higher accuracy, lesser complexity and faster computational time compared to LLBP technique.
Download Full PDF Version (Non-Commercial Use)