Face Recognition Based on Fusion of Multiple Masks Local Feature Sets Using Wavelet Transform

Although face recognition confronts several challenges, it has garnered considerable attention over the last two decades. A new proposal of local directional pattern (LDP) operator for face recognition has been introduced to overcome different challenges confronting face recognition systems. This new proposal is known as Multi-Mask Local Directional Pattern (MMLDP), which integrates the advantages of LDP operator depending on Robinson and Kirsch masks into one feature vector to give more discrimination information. Firstly, the LDP operator applies various masks to obtain the face image feature, and after that, methods such as Left-Right fusion, Right-Left fusion, Down-Up fusion, and Up-Down fusion are used to fuse the features. Finally, by employing a support vector machine (SVM), the classification of the feature ensues. The conducted experiments utilized the Yale database, and outcomes show that fusion of the features for various masks based on LDP improved performance with the best classification accuracy of 94.2857% compared to LDP and other conventional approaches for face recognition.