According to the Technology Review, the results of the Face Recognition Grand Challenge showed that machine recognition of human individuals has improved tenfold since 2002 and a hundredfold since 1995, and today the best face-recognition algorithms are even more accurate than most humans. National Institute of Science and Technology program manager for tests Jonathon Phillips says the improvement in accuracy is due to the development of high-resolution still images, 3D face-recognition algorithms and the recent availability of 3D sensors, which directly capture information about the shapes of faces.
Current recognition software also focuses more on distinctive features of a human face's surface, such as the curves of the eye sockets, nose and chin, where tissue and bone are most apparent and do not change over time. Carnegie Mellon Robotics Institute research Ralph Gross says 3D facial recognition can also recognize subjects from different viewing angles, up to 90 degrees, which was a problem before, possibly because most facial recognition technology was used for tasks involving ID cards and face scanners, which use full frontal faces of cooperative subjects under controlled lighting. High-resolution still images have also improved face-recognition technology with detailed skin-texture analysis.
Any patch of skin, called a skin print, can be captured as an image, broken into small blocks that algorithms can then measure, recording lines, pores and actual skin texture. Gross says skin-texture analysis is capable of identifying differences between identical twins, which is impossible using facial-recognition software alone.