1. INTRODUCTION
Damage caused by aging public infrastructure is a
a growing problem in Japan, and it is necessary to detect
internal damage in concrete structures before it surfaces.
One of the non-destructive inspection methods for
concrete is the electromagnetic radar method, which is
based on the amplitude and intensity of the received signal,
called B-mode (Figure1.)[1]. Since this method is
subjective and requires the skill of the person in charge of
the inspection, it is desirable to be able to estimate the
damage automatically and with high accuracy. Therefore,
the objective of this study is to establish an algorithm to
estimate the void thickness in concrete accurately and in
real-time.

2. ALGORITHM FOR ESTIMATING THICKNESS
As the void thickness decreases, it becomes impossible
to distinguish its peak due to the subtractive interference
caused by the upper and lower reflected waves. Therefore,
we focused on the frequency response because the
reflected wave has a frequency dependence depending on
the void thickness change[2]. Figure 2. shows the
spectrum for different void widths, and since the spectrum
also changes as the void thickness changes, we thought
that the void thickness could be estimated by performing
pattern matching of the frequency spectrum.
Figure 3. shows the results of applying the algorithm to
20mm and 110mm rectangular void. In addition to
spectral matching, information on the extreme values of
the time waveform and comparison of the spectral
the centroid of the theoretical value with that of the measured
value enables highly accurate estimation of the void
thickness at a lower computational cost.

3. CONCLUSIONS
By focusing on both the frequency response and time
the waveform of the reflected wave, I succeeded in
quantitatively estimating the void thickness even with a
very weak signal compared to rebar.