<p>Hyperspectral imaging (HSI) provides multiwavelength physiological sensing for standoff biometric detection; however, ambient light fluctuations limit the robustness of conventional systems. Here we introduce a lock-in camera-based HSI framework that rapidly modulates wavelength-specific illumination and synchronizes detection, enabling robust hyperspectral video reconstruction under varying ambient conditions. In photoplethysmography validation, the system estimates heart rate with errors below 3 bpm, outperforming conventional HSI, which typically exceeds 10 bpm. Using dual-wavelength illumination (660 nm, 940 nm), we further extract blood oxygen saturation (SpO₂) dynamics with a maximum error under 3% and a 2.7-fold improvement in mean accuracy under fluctuating light. We use machine learning models trained on the high-fidelity photoplethysmography signals to reconstruct blood pressure and electrocardiogram waveforms accurately. Our approach could offer a practical route for hyperspectral biosensing, advancing robust, multiparameter biometric detection for remote health assessment.</p>