Understanding the imaging process and role of illumination in finger vascular pattern recognition

Finger vein recognition as a promising biometric technique has drawn increasing attention from the biometrics community in recent years. Compared with other biometric traits, e.g., fingerprint, face, or iris, finger-vein characteristics cannot easily be copied, leave no traces, and are convenient to use. Obtaining good-quality vascular patterns in practice is quite challenging since the images may be blurred and have different intensity areas and varying contrast. In explaining the imaging process, the literature on finger-vein recognition does not go beyond the premise that the hemoglobin in the blood absorbs the light, while the other tissues scatter it. The explanation of image formation is usually not very detailed. Our research presents a model to obtain a better understanding of imaging of finger vein patterns and the role of illumination in the imaging process. It can be used to improve the acquisition device, resulting in improved image quality and recognition performance. Our research presented two types of models: a physical model and a qualitative theoretical model. The contributions of our research are: bone plays the main role in the imaging process, acting as a NIR light scatter, while blood and soft tissues absorb it, with hemoglobin in blood being the most absorbent; the proposed physical model leads to a better understanding of the illumination process, in particular around the joints; the presentation of a phantom finger that is capable of generating a realistic image with ground truth; a qualitative theory explains the projection of blood vessels close to the skin as shadows on the finger surface, allowing us to predict the impact of illumination widths and directions on finger vascular pattern imaging and recognition; narrower bundles of light do not affect which veins are visible, but they reduce the overexposure at finger boundaries and increase the quality of vascular pattern images; top illumination performs well because of a more uniform with less overexposure at finger boundaries, which results in more visible veins; illumination with various direction can be interoperable since they produce the same vascular pattern; the projected vein patterns are independent of illumination direction; a clean, controlled dataset that has a consistent position of the finger, minimizes longitudinal finger rotation.