Download Link 1 http://bias.csr.unibo.it/fvc2004/databases.asp
Download link 2 https://download.csdn.net/download/sereasuesue/12265216
Brief introduction
Part of the image
Using the following sensors / technology collected four different databases (DB1, DB2, DB3 and DB4):
DB1: The optical sensor of cross-matching "V300"
DB2: The optical sensor provides Digital Persona "U.are.U 4000"
DB3: Atmel's thermal scanning sensor "FingerChip FCD4B14CB"
DB4: synthetic fingerprint generation
NOTE: Due to the disturbance (see below) deliberately introduced, FVC2004 database significantly harder than FVC202 and FVC000 database. Therefore, we should neither error rates of different groups FVC game should not be considered fingerprint matching technology levels did not improve.
Student Science degree in computer courses University of Bologna (average 24 years) readily agreed to serve as volunteers to provide fingerprints:
Volunteers were randomized into three groups, 30 in each group; each associated with a database, associated with different fingerprint scanner (different objects in each database, in addition to a small amount of overlap in the two databases 5 volunteers) ;
Each volunteer was invited to show his / her own in three different meetings, each meeting at least two weeks interval;
CROSS collected for different fingers, finger placement to minimize the difference;
Image quality control is not performed, the platen sensor system cleaning is not performed;
In every class, each volunteer four fingers each have four impressions;
In the first course, the finger on the individual requirements of a slightly different vertical positions (1 prints and prints in 2), and low and high pressure alternately applied (3 prints and prints 4) of the sensor surface;
In the second class, it is required each finger exaggerated skin deformation (footprint 1 and 2) and a rotary (3 and 4);
In the third course, the finger is dried (imprint imprint 1 and 2) and the wet (4 prints and prints 3).
At the end of the data collection, each database were collected fingers 120 and each finger prints 12 (1440 impressions). As with previous versions, the size of each of the database used in the test is determined as the finger 110 width (w) of each finger and the depth (d) 8 th impressions (fingerprints a total of 880); collect some additional data to anti-collection / marking errors. From a finger (set B) 101 to 110 have been provided to the participant, for parameter adjustment algorithm before submission; then the finger from 1-100 number (set A) constituting a reference.
