Finger Scanning in Biometrics

It utilizes the distinctive features of the fingerprint to identify or verify the identity of individuals.

Components

Finger-scan systems comprises:

·         Image acquisition hardware

·         Image processing components

·         Matching components

·         Storage components

These components can be located within a single peripheral or standalone device, or may be spread among a peripheral device, a local PC and a central server.

The surface on which the finger is placed is called a platen, also referred to simply as a scanner. Platens can be made of various materials, including glass, plastic, silicon, and polymer. Proprietary coatings are used to prevent damage to the platen itself – a scratched platen reduces a device’s ability to acquire high quality fingerprints images.

A platen is one piece of a finger-scan module, the basic building block of a peripheral or standalone finger-scan device. A module normally contains a platen attached to a small printed circuit board, along with a standard connector that allows digitized information to be transmitted to the peripheral or standalone device. Many of today’s finger-scan modules are capable of performing all system functions – image acquisition, image processing, template generation, template matching, and template storage – within a very small package. These modules simply send match/no-match decisions to an external application or system.

Modules can be built into PC peripheral devices, embedded in keyboards, laptops, handheld devices, or PCMCIA cards; integrated into door control devices; or built into standalone terminals and readers. Eventually they may be commonly found in mobile phones, on the surface of smart cards, and in motorized vehicles. Depending on cost and system design requirements, peripherals designed for PC access can perform all functions internally. However, in most finger-scan systems, template generation and matching, being more processor-intensive activities, take place on a local PC or central server. Standalone and physical access devices are more likely to have built-in processing power, such that all of the biometric processes take place within the device itself.

The results of a finger-scan match are transmitted from the biometric system in the proper format for authentication to logical or physical access systems.

How finger-scan technology works

There are five stages involved in finger-scan verification and identification:

-          Fingerprint image acquisition

-          Image processing

-          Location of distinctive characteristics

-          Template creation

-          Template matching

Image acquisition

The first challenge facing a finger-scan system is to acquire a high-quality image of the fingerprint. Image quality is measured in dots per inch (dpi) – more dots per inch means a higher-resolution image. Today’s finger-scan peripherals can acquire images of 500 dpi, the standard for forensic-quality fingerprinting. The lowest DPI generally found in the market is in the 300-to-350 dpi range.

Image acquisition is a major challenge for finger-scan developers, because fingerprint quality can vary substantially from person to person and from finger to finger. Some populations are more likely than others to have faint or difficult-to-acquire fingerprints, whether due to wear and tear of physiological traits. In addition, environmental factors can impact image acquisition. In very cold weather, the oils normally found on a fingerprint (which makes for better imaging) dry up, such that fingerprints can appear faint. Users may need to press more firmly or even rub the finger into their opposite palm to ensure that a quality image is acquired.

For a finger-scan image to act as an effective enrollment, the center of the fingerprint must be placed on the platen. Many users unfamiliar with the technology will place their finger at an angle, such that only the upper portion of the finger print appears. This results in fewer distinctive features being located during enrollment and verification, reducing the likelihood of successful operations.

An additional factor in image acquisition that can affect a system’s accuracy and performance is the size of the platen. Over time, finger-scan vendors have developed smaller and smaller platens in order to manufacture small devices and to reduce costs. However, there may be a point of diminishing returns in terms of minimizing platen size. Very small platens acquire smaller portion of the fingerprint in a consistent fashion, leading to false rejections.

Image processing

Once a high-quality image is acquired, it must be converted to a usable format. Image processing subroutines eliminate gray areas from the image by converting the fingerprint image’s gray pixels to white and black, depending on their pitch. What results is a series of thick black ridges contrasted to white valleys. The ridges are then thinned from approximately 5 to 8 pixels in width down to a single pixel, for precise location of the feature.

Location of distinctive characteristics

There is a great deal of distinctive information on the average fingerprint-enough to enable large-scale searches using only one or two fingerprints. This information is fairly stable available throughout one’s life and differs from fingerprint to fingerprint, even for identical twins.

The fingerprint comprises ridges and valleys that form distinctive patterns, such as swirls, loops, and arches. Most fingerprints also have a core, a central point around which swirls, loops, or arches are curved. Deltas are points, normally at the lower left or right corner of the fingerprint, around which ridges are centered in a triangular shape.

Fingerprint ridges and valleys are characterized by discontinuities and irregularities known as minuate– these are the distinctive features on which most finger-scan technologies are based. There are many types of minuate, the most common being ridge settings (the point at which a ridge ends) and bifurcations (the points at which one ridges divides into two). Depending on the size of the platen and the quality of the vendor algorithm, a typical finger-scan image produce between 15 and 50 minuate – larger platens will acquire more of the fingerprint image, meaning that a greater number of minuate can be located.

Template Creation

Vendors utilize proprietary algorithms to map fingerprint minutiae. Information used when mapping minutiae can include the location and angle of a minutiae point, the type and quality of minutiae, and the distance and position of minutiae relative to the core. A user normally must place his or her enrollment fingerprint more than once during enrollment, so that the system can locate the most consistently generated minutiae.

Finger-scan images will normally have distortions and false minutiae that must be filtered out before template creation. For example, anomalies caused by scars(wounds), sweat, or dirt can appear as minutiae. Vendor algorithms scan images and eliminate features that simply seem to be in the wrong place, such as adjacent minutiae or a ridge crossing perpendicular to a series of other ridges. A large percentage of false minutiae are discarded in this process, ensuring that the template generated for enrollment or verification is an accurate reflection of the biometric data.

Template matching

Finger-scan templates can range in size from approximately 200 bytes to over 1,000 bytes – a very small amount of data by any measure. These templates cannot be manually read as anything resembling a fingerprint, and simply performing a bit-to-bit comparison of two finger-scan templates will not determine whether they are from the same person. Instead, vendor algorithms are required to process templates and to determine the correlation between the two.

Comparing enrollment and verification templates does not result in an exact match. The position of a minutiae point may change by a few pixels, some minutiae will differ from the enrollment template, and false minutiae may be seen as real. Also, the fingerprint will inevitably be placed at a slightly different angle. However, matching algorithms can account for these variations and allow for effective comparison of templates in which much of the underlying data may have changed.

There is no minimum number of minutiae necessary for two finger-scan templates to match. In some cases, the system may need to locate only a handful of minutiae in common to decide that two templates are match. Higher system thresholds will require that a higher percentage of the minutiae points match and can require more careful placement during verification. If a finger scan is deployed for 1: few identification against modest database as opposed to 1:1 verification, these thresholds will likely need to be increased. The most basic determinant of these thresholds will be whether the system is implemented for convenience or security.

Compiled By: Chaudhary Amit V.