Obtaining the image
Finger vein recognition works by shining invisible near-infrared light through the finger. The infrared light is absorbed by the haemoglobin of the blood in the veins. The result is an image of the unique pattern of veins which can be captured by a sensor placed below the finger.
Since haemoglobin strongly absorbs infrared light, the best images are obtained by shining light through the finger. In Figure 1 the light source is placed above the finger, with the sensor below. A practical device in this form-factor is a industrial standard reader, designed for PC logical access application.
For many applications, however, users would prefer simply to place their finger onto the device rather than into it, in which case illuminating from above becomes a limitation. The manufacturer has addressed this issue by developing a “side-illumination” technique. This technique combines the advantages of using transmitted light with the advantage of having an open, convenient device. A device in this form-factor is industrial standard embedded reader, which is suitable for incorporating into a wide range of applications.
To enable the finger vein device to cope with a wide variation in finger size and operating environment the light source intensity is adjusted adaptively. This enables the optimisation of image contrast and detail, and the minimisation of noise — an important issue due to the very high sensitivity of the image capture CCD.
The authentication process
There are four stages in finger vein authentication. These have analogues in most biometric techniques:
In stage 2 the finger vein image is normalised to accommodate geometric changes in position or angle of the finger. This is done by detecting the outline of the finger in the image, and rotating the entire image to normalise the slope of the outline.
Distinctive features of the finger vein pattern are extracted in stage 3. This is an essential step for eliminating variations in the captured data due to changes in body metabolism or imaging conditions. The result is a standard finger vein template of approximately 400 bytes which is suitable for passing to the matching algorithm.
In stage 4 the captured finger vein template is matched against a previously stored reference template. If a sufficiently close match is found then the user is authenticated.
In the example above, the reference template is stored in a smart card. If a capable enough smart card is used then the matching can take place on the card itself. This enhances security since the reference template never leaves the card. Alternatively, the reference templates can be stored in the finger vein device itself, on an attached PC, or elsewhere on the network.