Sensor Inter-operability / Platform Independence
With our algorithm you'll never need to re-enroll fingerprints if you change your sensor provider, the original template is always maintained and employable across any sensor from leading manufacturers.
Our algorithm provides sensor inter-operability. An original enrollment captured on a Harris / Authentec sensor (for example) is always maintained and employable in the event that verification is required on a sensor made by another provider like Veridicom, CSF Thomson, Siemens or ST MicroElectronics. Quite simply, users will never need to re-enroll if you decide to change your sensor provider or want to use multiple types of sensors based on particular needs.
A High Accuracy Rate
that helps Minimize Customer Support
Minutia systems cite accuracy rates of approximately 99%. Our algorithm gives
you 99.9% accuracy. When you consider false rejections, that's the difference
between 100 support calls for a user group of 10,000 or just one
support call for same size group.
Minutia systems cite accuracy rates of approximately 99%. Our algorithm gives you 99.9% accuracy. When you consider false rejections, that's the difference between 100 support calls for a user group of 10,000 or just one support call for same size group.
Reduced Silicon Costs
Unlike the minutia systems that were designed to work with large optical
sensors, our algorithm can maintain its accuracy even as the size of the image
captured continues to get substantially smaller in response to price pressures
on reducing silicon costs. The bottom line, silicon is expensive so we
designed our algorithm to be "data rich" in nature and require a
substantially smaller segment of the fingerprint for verification than
Fake Finger Detection
Unlike the minutia systems that were designed to work with large optical sensors, our algorithm can maintain its accuracy even as the size of the image captured continues to get substantially smaller in response to price pressures on reducing silicon costs. The bottom line, silicon is expensive so we designed our algorithm to be "data rich" in nature and require a substantially smaller segment of the fingerprint for verification than traditional systems.
Our Algorithm provides reliable detection of "Latex/Rubber" Fingers. The combination of our "Ridge Guide" which is incorporated in our sensor lock hardware device and the propriety image evaluation methods contained within our patented algorithm, provide us with the necessary information to dependably detect fake fingers.
Enrollments Templates that are Time Durable
Ridge recognition employs a revolutionary new "data rich" approach of comparing the complete ridge pattern of the finger versus traditional minutia based approaches which image the entire print, but only store and compare the minutia. With this approach we are able to clean up a fingerprint image, adjusting for the effects of dirty fingers, skin pores and cuts.
The Verification Process...
The fingerprint verification algorithm is the result of 20 years experience designing high-tech systems for the Department of Defense. The algorithm takes a two-dimensional imaging approach to pattern recognition for fingerprint verification. The approach is divided into two steps, enrollment and verification.
To enroll a new user, the system performs a process that creates a template. The template is made up of the user's ID number, information about their fingerprint and an image. A sensor performs the function of capturing fingerprint images. Once the image is captured, the Ridge algorithm begins applying filters to it. The first filter cleans up the image, adjusting for the effects of dirty fingers, skin pores, and cuts. The filtering is done in Fourier space, which allows a power spectral density function to be generated. This function tells Ridge important information about the dominant frequencies associated with the image. The filter is adaptive to each image, allowing the process to be automatic.
It is common that imaged fingers will have creases or be dirty, factors that hide the ridges of a fingerprint. To compensate, the algorithm performs a secondary filtering step that enhances and frequently restores the ridges in the image. The quality and content of the image is then assessed, if either are extremely low the enrollment will provide reduced accuracy. Low quality, which relates to the clarity of the ridge pattern, may be caused by lack of a fingerprint or severe damage to the print in the case of burns or scar tissue. A low content score can be caused by a print that lacks "complexity". The most unique areas of your fingerprint are termed cores. The locations of the most unique cores of a fingerprint are stored in the template. Cores are important to align images during the verification process. Enrollment templates are stored to be used in the verification process.
To perform verification, the Ridge algorithm collects an image of a candidate finger from a sensor; the image is pre-processed and matched against the corresponding enrollment template. The purpose of preprocessing is to minimize if not eliminate the differences in rotation, offset and distortion from the corresponding enrollment template.
The matching function incorporates the correlation of the candidate and template images, as well as noise and content weighting. The matching score is compared to a threshold that can be modified by the Administrator by changing the security level. If the verification score exceeds the threshold, the verification passes, if it does not the verification fails.