How facial recognition works
- Face algorithms rely on features of the face such as its geometry, appearance, and skin texture.
- Geometric information includes the relative position of and distance between key facial points such as the eyes, nose and chin.
- Appearance features examine the pixels across the whole face, giving more importance to areas which change the least over time.
- Where high resolution images are available, finer details of the skin can also be used, and can help differentiate between identical twins. Some of the most accurate face algorithms now use deep neural networks to learn these distinguishing facial features.
Face verification can be made even more reliable by constructing a 3D model of the face geometry; this works best with specialized infrared illuminators and cameras such as those now appearing in some flagship smartphones. This approach is more tolerant to the user moving, changing expression, or changing the orientation of their head.
Applications of fingerprint recognition
Facial matching is accurate and convenient and is often the preferred biometric selected by end-users. Almost every smartphone available on the market since 2010 has incorporated a camera capable of being used for biometric face capture. This makes facial recognition a great option that can be applied across most users.