White paper, September 2002
© 2002 Idex ASA. All rights reserved
Idex SmartFinger Algorithms
Sigmund Clausen
Idex ASA, Gamle Borgenvei 5, N-1385 Asker, Norway
This white paper describes the sensor specific algorithms accompanying the Idex
SmartFinger sensor.
In section 1 the image reconstruction algorithm is described and in section 2 we describe
the pointer/navigation algorithm
Idex can also provide object code of a fast and reliable fingerprint recognition algorithm.
However this algorithm is not sensor dependent and is not described here.
1. Image reconstruction: Why is it necessary?
Idex fingerprint sensor consists of a single line with 256 sensor elements comprising 500
dpi resolution. In order to capture an image of the whole fingerprint, the user must swipe
the finger across the sensor line. The swiping velocity and direction will typically vary
from one swipe to the other. This introduces unwanted degress of freedom into the
images like skewness, stretching and compression. These degress of freedom must be
removed prior to authentication and identification.
The purpose of the reconstruction algorithm is simply to remove any distortions in the
fingerprint images coming from a varying finger swiping speed and direction. A correctly
reconstructed image will be suitable for any fingerprint recognition algorithm taking into
account translational and rotational degrees of freedom. A repeatable geometrical
reconstruction of the images would imply that all distances within images captured from
the same finger do not vary by much. If this is the case the images can be aligned by
translation and rotation only. Almost any fingerprint matching algorithm relies on this
fact and it is therefore crucial for the biometric performance.
To put it simple, the reconstruction algorithm calculates the distance travelled by the
finger “on the fly” and adds rows to the reconstructed image at every time instant the
finger has moved 50 μm, ensuring 500 dpi resolution in both the pulling direction and
along the sensor line. The examples above illustrate the purpose of the reconstruction
algorithm. After image reconstruction the two images can be aligned and compared
directly by allowing for translational and rotational degrees of freedom. Image-
skewness, stretching and compression is removed by the reconstruction algorithm.
评论1