Optical Skin Detection for 3D Face Authentication Against Spoof Masks

The patent segments the face recognition process into distinct functional modules: illumination pattern projection unit, camera unit, processing unit for beam profile analysis, and authentication unit. This segmentation allows each module to perform specific tasks efficiently, reducing overall computational burden while maintaining accuracy through specialized processing at each stage.

The patent changes the illumination parameters by projecting patterns with specific beam profiles (Gaussian, uniform, or other distributed patterns) onto the face. By analyzing how different illumination patterns reflect off the face and are captured by the camera, the system creates distinctive signatures that enable accurate recognition with lower computational complexity compared to traditional deep learning approaches.

The patent replaces traditional mechanical/video-based PAD approaches with an optical field-based system. Instead of analyzing temporal variations in video frames or complex 3D geometric analysis, the system projects illumination patterns and analyzes the reflected optical field characteristics. This substitution enables faster processing while maintaining robustness against spoof attacks through physical optical property analysis.

The system uses periodic illumination pattern projection and analysis cycles. By projecting patterns at different positions and analyzing the reflected light in a systematic sequence, the system efficiently gathers sufficient information for authentication decisions without requiring prolonged processing time. The periodic nature of the measurement process enables optimized timing for secure and fast authentication.