Verifikasi Wajah Berbasis Data Citra Menggunakan Kombinasi Fitur Tekstur
Nurdana Ahmad Fadil, Wahyono, S.Kom., Ph.D.
2024 | Tesis | S2 Ilmu Komputer
The face is the part of the body located at the front of the human head, covering the area of the forehead, nose, eyebrows, cheeks, and chin. Each individual has distinct facial features, making the face a significant object in the field of biometrics. Biometrics is an automated method that can identify a person based on their physical features, such as eyes, fingerprints, signatures, and voice.
The process of face verification involves several steps, starting with initial preprocessing, which includes converting the image to grayscale and resizing the image to uniform dimensions of 140, 160, and 180 pixels. Following this, texture-based feature extraction is performed using three methods: LBP (Local Binary Pattern), LPQ (Local Phase Quantization), and BSIF (Binarized Statistical Image Features). Additionally, feature dimension concatenation and difference extraction from image pairs, known as differentiation, are conducted using combined methods such as LBP+LPQ, LBP+BSIF, and LPQ+BSIF.
Next, dimensionality reduction is carried out using PCA (Principal Component Analysis) with component sizes of 80, 90, 100, 110, and 120. The extracted and reduced features are then classified using KNN (K-Nearest Neighbor) and SVM (Support Vector Machine). The highest accuracy was obtained using the combined LBP+LPQ method with a 140-pixel image size and concatenated features, resulting in an accuracy of 0.95 with KNN and 0.97 with SVM.
Kata Kunci : Verifikasi Wajah, Ekstraksi Fitur, LBP, LPQ, BSIF, KNN, SVM