The two most popular biometrics in use today are iris and fingerprint. With the appropriate hardware and environmental settings, these biometrics are both accurate and robust. Their primary drawback has been the requirement of additional hardware to make them work. Iris biometrics requires near infrared light and image sensor to extract the features needed for matching. Likewise, the fingerprint modality requires specialized capacitive, RF, or optical add-on hardware which are not present on cell phones.
When addressing the mobile marketplace, one must assume that only the mobile device itself can be used to capture the biometric of interest. Thus, Iris and fingerprint are not suitable for widespread adoption amongst mobile users.
Facial recognition is feasible on mobile devices. Google, Apple and many others have demonstrated this capability. However, the consistency and accuracy of facial recognition as a stand alone biometric is too poor to use reliably.
Facial recognition does allow for an interesting blending of Eye Vein and Facial Recognition Biometrics. Both leverage standard cameras to image the same general area. If one captures an image used for Eye Vein Biometrics, they have by default also captured the face as well. Fusing the results of these two modalities allow for a more robust and redunt verification system.
Voice recognition, like face, is a solid biometric in the right environment. Variability in the voice itself as well as background noise has been challenging in mobile and unconstrained environments.