 |
As a consequence of recent events, public safety has become a centerpiece of our global society. Motorola’s relevant expertise in fingerprint identification, based on more than a 30-year history of successful development and worldwide deployment, constitutes one element of the larger public safety enterprise. Work is under way to extend this technology by including other biometric and investigative information sources for delivery not only to desktops but also wirelessly through Motorola’s seamless mobility technology. While this evolution continues, the demand for accurate and rapid identification of fingerprints is rising in various civil scenarios. In response, Motorola has been directing serious efforts to extend the state of the art by exploring advanced image processing, feature extraction, and other analytical methods in order to achieve greater accuracy under the added requirement of improved speed. For over four years, Motorola’s Biometrics Advanced Technology Group has been using Mathematica to efficiently formulate and evaluate analytical techniques applicable to various aspects of fingerprint identification. The computation of a fingerprint’s image quality map, segmentation, and ridge-flow direction image, the automatic detection of its singularities, the characterization and compensation of image distortion, and minutiae registration, matching, and scoring are a few examples. Mathematica is being used to design interactive applications for gathering ground truth data to support subsequent investigations and to evaluate the quality of automatically detected data. Mathematica movies have also provided invaluable visual insight into the functionality of underlying algorithms. This and much more is being done on a backdrop of an experimental system containing a large variety of fingerprint data, organized along with a collection of Mathematica notebooks in a structured multidirectory Mac OS X system. Periodic reports of work accomplished is recorded in informative, executable Mathematica notebooks and distributed in PDF format. While much progress has been made to date, there are some very difficult problems that remain to be addressed more successfully-image enhancement, background isolation, separation of overlapped prints, maximum likelihood exploitation of detected features having limited quality and observability, analytic image continuation based on distortion models or knowledge of complex-plane singularities, classification, and so on. These and other related open issues are most clearly and efficiently addressed in Mathematica, the natural platform for the task. But Mathematica’s potential goes beyond its apparent function as an analytical testbed. With the upcoming release of Mathematica 6, the possibility of developing useful interactive fingerprint applications, such as a Latent Examiner’s Workstation, is just over the horizon. Even more intriguing is the very real possibility that, by virtue of its existing SQL database interface, Mathematica could support the essential functionality of an Automated Fingerprint Identification System, if not a more extensive Biometrics Identification System, not only as a platform for mathematically disciplined development but also as one capable of limited deployment.
|
|
For the newest resources, visit Wolfram Repositories and Archives »
