Face Spoofing Detection Using Hybrid Binary Statistical Image Features-Gaussian Filter

Face presentation attacks threaten the dependability of facial biometric authentication, particularly when conventional RGB cameras are exposed to print and replay artefacts. Existing texture descriptors are computationally attractive but remain sensitive to noise, illumination variation, and image degradation, whereas many deep architectures require substantial data and processing resources. Prior studies establish the value of handcrafted texture features, Binary Statistical Image Features (BSIF), auxiliary-supervised networks, and foundation-model approaches, but the reviewed evidence does not explicitly assess Gaussian linear filtering as a face-image preprocessing stage before BSIF extraction. This study adopted is structured into sequential and interconnected phases were integrates Gaussian linear filter with sigma 1.0, eight-bit BSIF coding, a normalised 256-bin histogram, feature standardisation, and support vector classifier (SVC). Experiments use OULU-NPU, with five-fold validation on SiW and CASIA-FASD. On the reported OULU experiment, the system achieved 87.28% accuracy, 65.83% precision, 76.70% recall, a 70.85% F1-score, and 93.39% ROC-AUC, compared with 73.50% accuracy and 0.724 ROC-AUC for raw-pixel SVC. Mean accuracy was 83.69% on SiW and 95.05% on CASIA-FASD. Gaussian filter- BSIF features provide a compact and discriminative representation for software-only face presentation attack detection. The work demonstrates a lightweight preprocessing-texture-classification pipeline and reports both predictive and computational measures. Future studies should enhance the robustness, scalability, adaptability, and efficiency of face presentation attack detection by integrating advanced preprocessing techniques, multimodal biometrics, lightweight artificial intelligence models, and adaptive learning strategies.