Page 75 - Riem-Vol6_nº3

420 IBRACON Structures and Materials Journal • 2013 • vol. 6 • nº 3
Monitoring of crack opening in masonry and concrete using mobile platform
On the measuring comparison, the values of the proposed meth-
od, that resulted the followinggraphicswerecollected, processed
and presented to the user directly on the mobile phone’s screen.
Figure 8 presents a graphic illustration external to the mobile
phone, with the increase rates of the crack opening registered
by the proposed system on three different specimens. The
measuring values were obtained on 22 gaps of advancement
application axis of compression of the testing machine. The
measuring was accomplished on the three specimens after the
appearance of the first crack.
The other results analysis in this section refers to the speci-
men number three. This specimen was chosen due to its best
visual positioning of the cracks disposed in the testing ma-
chine, because the specimen is not withdrawn after the begin-
ning of the compression test. Another factor was the improve-
ment of the comparison with the proposed method against the
conventional method, due to the bigger amplitude of the crack
opening observed.
To validate the proposed method, the captured images were
submitted to a comparative test with off-line processing on PC.
In that comparison it was employed a toolbox image process-
ing (Matlab) and a pixels counting algorithm. Figure 9 illus-
trates the results comparision on PC image processing mea-
suring and the proposed method.
The proposed method followed the referential measuring of the
processed algorithm on the PC software and the difference
between the methods represents a value below 1.4%.
On a periodic inspection, the technician verifies the increase
or stabilization of the crack in relation to the period of time
between monitoring, without the metric relation between the
crack on masonry and the image. It is only considered the
difference of occupation of the crack during the observed im-
ages. Figure 10 presents the real mobile phone’s screens on
the proposed method.
Figure 10 (c) presents a captured image of the masonry struc-
ture with different dimensions on the crack opening. Figure
10 (d) presents the binary image with a total result of 307200
pixels, and 16.198% of crack occupation (black pixels) on the
monitored image. Regardless the length, index and opening
dimension, all the visible by eye cracks were processed by the
proposed method.
Figure 11 presents a comparative analysis between measures
presented by the use of a graduated scale, a digital pachym-
eter and the proposed method. Due to the difference between
the proposed method and the measures obtained with the use
of a graduated scale (9% and 34%, respectively), the values
adopted as reference on the conventional method were ob-
tained with a digital pachymeter, with measuring variations be-
tween 4.2% and 4.9%.
The same region of the specimen presenting crack opening
was utilized on all measures, regardless the measuring meth-
od. The use of the scale and pachymeter demanded more time
to obtain the conventional measures, due to the two measures
of the two technicians on each instrument. The successive
measures were necessary to obtain the corresponding aver-
ages for each phase of the cracks evolution, and are illustrated
in Figure 11. In total seven cracks were observed during the
period of the specimen compression on the testing machine.
Along that procedure of compression, significant differences
on the dimensions of the crack opening were observed.
Different from the conventional technique, that demands extra
calculations from the manual readings, the proposed method
the percentual of crack occupations as the images were cap-
tured. The 0.5 mm resolution of the graduated scale, associ-
ated to the difficulty of preview (scale reading and positioning
on the crack) contributed to produce measures with higher er-
Figure 8 – Area % occupied by cracks on the
specimen obtained in the proposed
method with the mobile phone device.
There were 22 average measurements on each one
of three specimen during the comparison experiment
Figure 9 – Measurement comparison
between the images in the proposed
method and the images processing in PC.
Total were 22 average measurements