Cultural Heritage Applications: Archaeological Ceramics and Building Restoration - On Statistical Pattern Recognition in Independent Component Analysis Mixture Modelling - page 145

Table 6.6 Equipment setup
Ultrasound equipment setup Acquisition equipment setup

Ultrasound
equipment

Matec PR5000 Acquisition

equipment

Oscilloscope TDS3012
Tektronix

Transducers 1 MHz (application (1) Sampling

frequency

10 and 250 MHz

5 MHz (application (2)

Pulse width 0.9 ls Sample number 10,000

Pulse amplitude 80 % Observation

time

1 ms and 40 ls

Analog filter 200 kHz-2.25 MHz

(consolidation diagnosis)

Vertical
resolution

16 bits

1 MHz-7 MHz (layer
determination)

Excitation
signal

Tone burst 1 and 5 MHz Dynamic range 2.5V

Operation mode Pulse-echo Average 64 acquisitions

Amplifier gain 65 dB PC connection GPIB

the thickness layer profile (application 2) was 5 MHz. This last transducer was
selected because we were interested in obtaining information of the superficial
layers.
The equipment setup used for NDT of the historical building is described in
Table 6.6 (an outline of equipment connections is in Fig. 6.7 ): The Mixca algo-
rithm described in Chap. 3 , which was configured to estimate one ICA, was used
(see Appendix A). The mixture matrix obtained by ICA was used to separate the
information concerning the sinusoidal phenomena.

6.2.3 Diagnosis of the Material Consolidation Status

BSS by ICA was selected for this application because, contrary to the classic
spectral analysis techniques [ 32 ], BSS is an unsupervised method that does not
require any estimation of the noise autocorrelation matrix in data corrupted by the
sinusoidal interference, considerations on the kind of noise, or model assumptions
such as the filter order in model-based methods.
Figure 6.8 shows the B-Scan estimated by signal power using a conventional
non-stationary analysis applying a moving window over the 12 ultrasonic recorded
signals. Figure 6.8 a shows two clearly differentiated zones; the first zone corre-
sponds to the consolidated zone (low level of signal) and the second zone corre-
sponds to the non-consolidated zone (high level of signal). The signal penetrates
well into the wall at the consolidated zone and is attenuated before reflecting any
kind of signal. Conversely, the signal level is increased in a non-consolidated zone
due to multiple reflections of the ultrasonic pulse (see Fig. 6.8 b).