Instructions

U.are.U SDK - Developer Guide 11
Developing Applications
THIS CHAPTER DESCRIBES THE PROCESS OF DEVELOPING FINGERPRINT RECOGNITION APPLICATIONS.
This overview covers the following topics.
• How fingerprint recognition works
• Data structures and data flows among components
• Typical workflows
• Design issues and tradeoffs.
The JavaPOS and OPOS APIs conform to the UPOS specification and therefore do not have exactly the same data
structures and workflow as the other APIs in the U.are.U SDK. See the following chapters,
“JavaPOS” and “OPOS” for
details.
How Fingerprint Recognition Works
Fingerprint recognition works in two stages:
• First, users are enrolled with the system--their fingerprints are captured and stored in a database.
• Next, when a person needs to be given access (e.g., to open a door or to log in to a computer), they simply scan
their finger on the fingerprint reader.
In terms of application development, this typically requires the developer to build the following components:
• An application that enrolls users:
• Captures multiple fingerprints for at least two fingers from a fingerprint reader.
• Checks image quality to ensure that a good quality scan is obtained.
• Extracts the fingerprint minutiae.
• Saves the fingerprint images and/or minutiae in a database.
• A service(s)/application(s) that identifies/verifies people:
• Captures a fingerprint from a fingerprint reader.
• Extracts the fingerprint minutiae.
• Compares fingerprint with enrolled fingerprints to identify a user from a list or verify a specific user.
This SDK provides fingerprint capture, extraction, enrollment and identification/verification functions to help you
develop these components.
Understanding the Data Flow
When building a fingerprint recognition application, the data flow consists of:
1 Capture a Fingerprint Image (scan) from the fingerprint reader. The resulting Fingerprint Image Data (FID) con-
tains one or more fingerprint images, called a Fingerprint Image Views (FIVs). A typical FID for fingerprint recog-
nition applications contains only one FIV but we also support multiple views (e.g., if there are multiple fingers from
one individual or multiple images from a single finger stored in a single FID).
Each FIV (fingerprint) is approximately 140K in size.
2 Extract the fingerprint features. During extraction, Fingerprint Minutiae Data (FMD) is created, with each
fingerprint stored in a Fingerprint Minutiae View (FMV) in the FMD. An FMV in an FMD takes no more than 1.5K
(maybe less depending on the fingerprint). FMDs are used for identifying users in a collection and verifying
specific users.
The ANSI and ISO standards permit multiple views but the U.are.U SDK creates only single-view FIDs and FMDs.