The identity comparison methods designed for images
captured by classical touch-based sensors cannot be directly
applied on touch-less fingerprint images. For this reason, the
biometric recognition systems in the literature designed for
touch-less fingerprint images apply image processing algorithms
for improving the image quality and then compute
the match-score between two images by applying classical
algorithms. It is possible to distinguish methods that computes
three-dimensional models [1, 2], fingerprint images
obtained by the mosaic of a set of images [3], and methods
based on single fingerprint images [4, 5]. The methods based
on multiple finger images estimate rolled-equivalent images,
and then apply classical identity comparison methods. These
methods obtain good performances in terms of accuracy but
are computationally expansive. The methods based on single
images apply image enhancement techniques, and then use
classical identity comparison methods. Usually, the required
computational time is less than the time required by the
methods based on three-dimensional models, but the obtained
accuracy is lower.