Fingerprint verification technology?

1. Fingerprint verification technology

Fingerprint verification technology
Fingerprints can be captured in different ways. Current techniques include optical, ultrasound, or technologies based on semiconductor chips(CMOS sensor). Optical method is the pioneering technology of the fingerprint verification and widely being used around the world.

Types of sensor Optical Sensor

All optics-based fingerprint systems translate the illuminated images of fingerprints into digital code for further software processing like enrollment (fingerprint registration) and verification (authentication of registered users). Algorithm of this technology varies from one company to another however, it is very stable and reliable

2. Semiconductor Sensor

There are three ways to capture fingerprint image using semiconductor sensor. One is capturing image of fingerprint that is laid on the chip surface, another is using e-field or ultrasound to read the fingerprint image. Some critical factors that contribute to high error rate of semiconductor sensor like ESD, moisture, dirt and oil have been dissolved. Also, durability of the chip is also an issue as users are touching it directly however, it also has been dissolved by special coating and protected cover or design.

3. Algorortm

Extraction & Comparison Process
The feature extraction process is performed in two stages. The first stage of is to capture a grayscale image of the fingerprint from the finger itself. The image is then processed down to a binary "skeleton" of the original image with one-pixel wide ridgelines. In the second stage of the process, the feature extraction algorithms are
applied to this skeletal image and the measurements that comprise the minutiae template are taken. Then, the templates are stored. When a new fingerprint is read, the algorithm compares it with the stored one.

Comparison Method

1:N

Identification(1:N) vs. Verification(1:1) : The Difference Identification is the process of selecting one person's characteristics from a group of records. Called a "one-to-many" search, the question put to the system is "Do you know this person?" The algorithm searches the database and returns a result of likely candidates.Verification, on the other hand, occurs when a person makes a claim to a specific identity. Called a "one-to-one" match, the question put to the system is "Is this person who he claims to be?" The automated system compares the individual's measured characteristics against a previously registered record to determine whether or not the match is valid.