|
|
|
|
|
|
 |
 |
|
Implementation and Visualization of Fingerprint Minutiae Matching with Mathematica: Shortcomings and Options for Improving Accuracy
|
|
|
|
|
|
|
|
|
|
|
|
2006 Wolfram Technology Conference
|
|
|
|
|
|
Champaign IL
|
|
|
|
|
|
With its dynamic elements and associated interactive graphics, Mathematica 6 offers an important capability to control, observe, and evaluate the dynamic behavior of fingerprint matching algorithms. Using the Manipulate function, the classical minutiae matcher (MM) can now be exercized visually, imparting thus an intuitive understanding of the underlying registration and matching constituent stages. Detected minutiae from a search print are superimposed over those of a file print, rotated at fixed angular increments over a prescribed azimuth range, registered by identifying quantized mutual subsets of translational similarity, matched by effecting a derived translation at full-pixel resolution, and producing a score based on a reasonable Euclidean measure. Through controls provided with the interactive graphics, one can choose to observe the state of MM, at either the registration or the matching stage, for a given azimuth angle. Alternatively, the behavior of each of the stages may be examined dynamically over the specified azimuth scan range.
While MM is expected to be reliable in an identification setting where minutiae detection consistency can be controlled, it is less dependable in situations where image quality and relative distortion between search and file prints tend to diminish the reliance on Euclidean measures. In the latter case, some of the consequent loss in accuracy could be recovered with probabilistic and transform-based methods. The likelihood of a correct match could be increased by weighting individual minutiae-matching measures according to the image quality maps of the two fingerprints being compared. The reliability of the match could also be enhanced by taking into account the ridge structure among corresponding pairs of minutiae from the two fingerprints. The latter method, which has been shown to significantly improve MM matching performance in practice [1,2], is explained in some detail.
[1] URL for Gray Scale Matcher http://www.motorola.com/governmentandenterprise/contentdir/en_US/Files/PressReleases/PR_2005Jun07_Hellenic_Police.pdf#search=%22Gray%20Scale%20Matcher%22 [2] URL Gray Scale Patent Pending http://www.google.com/search?q=%22Sam+M.+Daniel%22&hl=en&lr=&start=10&sa=N
|
|
|
|
|
|
|
|
|
|
|
|
fingerprints, Minutiae Matcher, matched minutiae pairs, cross section correlation technique, latent profile, bandpass filtering, registration, matching
|
|
|
 |
|
|
| Daniel-WTC06.nb (13.8 MB) - Mathematica Notebook [for Mathematica 6.0] | | datafiles-part1.zip (8.4 MB) - Data files - part 1 | | datafiles-part2.zip (7.9 MB) - Data files - part 2 | | supportfiles.zip (2.2 MB) - Documentation files and additional supporting files |
|
|
|
|
|
|
|
| | | | | |
|
For the newest resources, visit Wolfram Repositories and Archives »
