An Optimized Feature Extraction Method (LBP+WOA) for Anti-Face Spoofing

Recognition systems are highly susceptible to spoofing attacks such as video replays, printed photos, 3D masks and authentication security systems. To address these challenges, it is proposed to have an optimized feature extraction technique that combines Local Binary Patterns (LBP) and the Whale Optimization Algorithm (WOA) to improve various problems faced in spoofing detection. LBP is an efficient technique that is a texture-based feature extraction approach which uses micro-texture changes to discriminate between real and artificial faces. But traditional LBP approaches frequently produce duplicate or irrelevant features, which can compromise classification performance. To improve parameters such as efficiency, WOA is used to optimize feature selection by reducing dimensionality while keeping critical discriminative information. The proposed LBP+WOA technique is tested on benchmark face anti-spoofing datasets which include the NUAA Impostor Database and shown higher detection accuracy and lower false acceptance rates than existing LBP-based techniques. The hybrid of LBP's robust texture analysis with WOA's intelligent feature selection yields a lightweight and efficient model suitable for real-time anti-spoofing tasks. This technique protects face recognition systems against spoofing attempts while being computationally efficient. In the future, this research work will focus on deep learning-based hybrid strategies that improve performance against sophisticated threats such as adversarial manipulation and deep fake-based spoof.