Life recognition based on color variations in fingerprint images of live and artificial fingers

摘要

Life-related signal detected from a finger-tip can be used to prevent frauds by finger-replicas. A finger deformation induces blood movement and the light scattered inside the finger carries this life-related information. We propose to look at the changes in color and luminance extracted from the central part of the fingerprint images. In experiments, we examined input actions taken by more than 30 participants as well as six replicas made of various materials. For the live fingers, chromaticity coordinates x and luminance Y showed a relatively large hysteresis as a function of the area of the fingerprint images. This hysteresis was smaller in case of the replicas. Based on this fact, we were able to define indices and criteria for life recognition so that all the replicas were rejected while the most live fingers were accepted. However, the recognition was not perfect due to the small hysteresis shown by some replicas. Improvements in the input hardware and the algorithms for life-related information extraction need to be addressed in future.