Case Studies - Measuring the User Experience: Collecting, Analyzing, and Presenting Usability Metrics - page 248

EFFECTIVENESS
Three dependent variables are related to the effec-
tiveness of the self-capture system: number of par-
ticipants that completed the tasks (task completion
rate), errors, and quality of the fingerprint image.

(a) Start of the
capture process

(b) RS sym bol (c) Scanning

area with RS
overlay

TASK COMPLETION RATE
Overall, 37 out of 40 (92.5%) participants for Task
1 and 39 out of 39 (100%) participants for Task 2
completed the self-capture tasks successfully by fol-
lowing the on-screen instructions without assis-
tance or prompts from the test administrator.

(d) Correction
example -
“Rotate
counterclockwise”

(e) Example of
an acceptable
RS

(f) RT symbol

ERRORS
The average number of hand-positioning correc-
tions (errors) for each fingerprint is shown in Table
10.1 . As shown in Table 10.2 , errors can be classified
into four categories.

(g) Scanning
area with RT
overlay

(h) Correction
example -
“Move up”

(i) LS symbol

Figure 10.8 shows the seven most common
errors. For slaps (both right and left hands), not
enough pressure applied was the most common
error: RS: 214, LS: 101. There were many occurrences
where not all four fingers were detected at the same
time, indicating that participants had some difficul-
ties placing their fingers evenly on the scanner: RS:
132, LS: 51.

(j) Scanning
area with LS
overlay

(k) Correction
example -
“Move right”

(l) LT symbol

QUALITY OF THE FINGERPRINT IMAGE
We used NIST fingerprint imaging software to com-
pute the NFIQ ( Tabassi et al., 2004 ) score for each
finger. NFIQ scores range from 1 (highest quality)
to 5 (lowest quality). The medians of individual
NFIQ scores are shown in Figure 10.9 .

(m) Scanning
area with LT
overlay

(n) Correction
example -
“Press more”

(o) End of the
capture process

Figure 10.7 Overlay condition—fingerprint self-capture
process with examples. 2

As there is not yet consensus in the biometrics community on how to deter-
mine the quality of a slap image, we used a proposed quality scoring method
under consideration by US-VISIT to assess the overall quality of the images. A
slap is accepted if the index finger and middle finger have an NFIQ value of 1 or
2 and the ring finger and little finger have an NFIQ score of 1, 2, or 3. A thumb
is accepted if it has an NFIQ value of 1 or 2. The results of applying the criteria

2 The fingerprint images were blurred to prevent possible identification of the participants.