Matched Filterss

Introduction to matched filters
John C. Bancroft
ABSTRACT
Matched filters are a basic tool in electrical engineering for extracting known wavelets
from a signal that has been contaminated by noise. This is accomplished by cross-
correlating the signal with the wavelet. The cross-correlation of the vibroseis sweep (a
wavelet) with a recorded seismic signal is one geophysical application. Another
geophysical application is found in Kirchhoff migrations, where the summing of energy
in diffraction patterns is equivalent to a two-dimensional cross-correlation.
The basic concepts of matched filters are presented with figures illustrating the
applications in one and two dimensions.
INTRODUCTION
1D model for matched filtering

Matched filtering is a process for detecting a known piece of signal or wavelet that is
embedded in noise. The filter will maximize the signal to noise ratio (SNR) of the signal
being detected with respect to the noise.
Consider the model in Figure 1 where the input signal is s(t) and the noise, n(t). The
objective is to design a filter, h(t), that maximizes the SNR of the output, y(t).
Signal s(t)
Filter h(t) y(t) = sh(t) +nh(t)
Noise n(t)
FIG. 1: Basic filter
If the input signal, s(t), is a wavelet, w(t), and n(t) is white noise, then matched filter
theory states the maximum SNR at the output will occur when the filter has an impulse
response that is the time-reverse of the input wavelet. Note that the convolution of the
time-reversed wavelet is identical to cross-correlation of the wavelet with the wavelet
(autocorrelation) in the input signal. When the wavelet is of length, T, then the matched
filter is defined by:
h ( t ) = w (T − t ) .
This result is derived in many signal processing texts such as Ziemer and Tranter
(1988) and Lathi (1968), and will not be derived in this paper. Essentially, the least-
squares principle is used to maximize the output signal energy with respect to the output
CREWES Research Report — Volume 14 (2002) 1

Bancroft
noise energy. The matched filter improves the SNR by reducing the noise’s spectral
bandwidth to that of the wavelet, and in addition, reduces the noise within the wavelet’s
bandwidth by the shape of the wavelet’s spectrum.
The duration of the wavelet can be small, as used in radar or sonar, and it can be much
larger when used with the vibroseis source with the acquisition of seismic data. Other
applications may involve the detection of weak signals from satellite transmissions, or the
detection of military equipment from visual images. In addition, Kirchhoff migration is a
form of 2D or 3D matched filtering that estimates the location, size, and shape of
scattered or diffraction energy.
EXAMPLES
Example with a short wavelet

The first example is a short wavelet added to noisy traces. In this situation, the signal-
to-noise ratio (SNR) is defined by taking the peak amplitude of the wavelet and dividing
it by the root-mean-squared (RMS) measure of the noise (standard deviation).
Various amplitudes of the wavelet in Figure 2 are added to ten noisy traces and are
illustrated in Figure 3. The SNR of these traces varies from 2.0 at the top to 0.2 at the
bottom as indicated on the left side of the figure.
FIG. 2: Wavelet, only first 100 ms (50 points) used.
The cross-correlation of the known wavelet in Figure 2 with the traces in Figure 3
produces the corresponding matched-filter traces in Figure 4. Note the improved SNR of
the traces, and that even the bottom trace with an initial SNR of 0.2 demonstrates a higher
probability of detection.
2 CREWES Research Report — Volume 14 (2002)

SNR
1/3
2/5
1/3
2/7
2/9
1/5
FIG. 3: Ten noisy traces and wavelets with a SNR that varies from 2.0 to 0.2.
FIG. 4: Results of a matched filter from cross-correlating the wavelet in Figure 2 with the noisy
signals in Figure 3,

Bancroft
Figure 4 illustrated the attenuation of the noise and the preservation of the wavelet
energy. The shape of the wavelet, however, has been modified to zero-phase,
corresponding to the cross-correlation of the wavelets being an auto-correlation. The peak
of the zero-phase wavelet identifies the beginning of the initial wavelet.
Example with a longer chirp wavelet

The above experiment is repeated with a chirp wavelet shown in Figure 5 and the
noisy signals in Figure 6. This wavelet is longer and contains higher frequencies than the
short wavelet. These higher frequencies also define the shape of the matched filter and
will allow higher frequencies of the noise to pass as observed in the matched-filter result
in Figure 7.
FIG. 5: Chirp wavelet.
FIG. 6: Noisy traces and chirp wavelet.

FIG. 7: Match-filtered result of the noisy chirps.
With a larger energy wavelet, there is more energy in the cross-correlation, and better
detection. Note that the input SNRs was the same as in the shorter wavelet example.
2D data
The same principles apply to detecting a 2D wavelet in a 2D signal. Such applications
are used to detect potential military equipment in video or optical images. In geophysics,
it can be used to detect seismic diffractions in a process we know as seismic migration.
Note that the matched filter concept tells us that we need to cross-correlate with the same
size and amplitude of the “object” being detected.
Consider the noisy section, the 2D wavelet, and the matched filter result in Figure 8.
The noisy section in (a) has 100 by 100 samples, and contains one barely detectable
wavelet. The wavelet in (b) has 21 by 21 samples and is much narrower when inserted
into the grid of the noisy section. The cross-correlation in (c) shows a significant
improvement in the matched filter result.

Bancroft
a) b)
c)
FIG. 8: Example of 2D matched filtering with a) the noisy section containing the wavelet and b)
matched filter.

COMMENTS
In the cross-correlation process, the amplitudes of the input wavelets or events are
squared and will represent some form of energy.
Seismic migration is similar to the detection process. Both processes try to collapse
the energy in a diffraction to the smallest possible point with the highest energy.
In migration, the diffraction stack algorithm cross-correlates a time and spatially

varying diffraction with a seismic section. For maximum detection, the Kirchhoff
operator should match the shape and amplitude of the diffractions in the section. Note
that the amplitudes on the migrated section would now represent the energy, and in the
exploding reflector model, the energy is proportional to the reflectivity of the interfaces.
This heuristic or ad hoc approach to migration coincides with the Kirchhoff integral
solution to the wave equation.
The reflectivity coefficients vary with the angles of incidence, reflection and
transmission. The inclusion of these amplitude factors in the weighting of the summation
diffraction can improve the signal-to-noise ratio of specific targets in the migrated
section. This has been demonstrated in the migration of P-P Class 2 AVO data and in the
migration of converted-wave data (Beckett and Bancroft, 2002).
In Class 2 AVO, the reflection coefficient varies from a positive to a negative value,
and summing (or stacking) these amplitudes tends to give a very small amplitude, relative
to the surrounding reflectors. Including these amplitude weightings in the prestack
migration operator (the diffraction shape) will boost the amplitude of the Class 2 event,
and attenuate the amplitudes of the data with a constant amplitude with offset. This
filtering should then aid in the detection of these events.
Converted-wave reflection amplitudes tend to zero at zero-offset and increase in

magnitude with offset. In addition, the noise toward zero-offset may be large and
contribute significant noise to the migrated section. Using the matched filter concept to
bias the migration operator with the same weighting will tend to attenuate the near zero-
offset noise to produce a migrated section with less noise.
Authors such as Claerbout (1992), point out that the inclusion of these AVO
weightings in prestack migrations will prevent further evaluation of subsequent AVO
parameters in any following prestack measurements.
CONCLUSION
Matched filters are designed to extract the maximum SNR of a signal that is buried in
noise. Some applications in geophysics are in the cross-correlation of vibroseis data, and
in Kirchhoff migration.
ACKNOWLEDGEMENTS
We acknowledge NSERC and the CREWES sponsors for their continued support.

Bancroft
REFERENCES
Beckett, J. and Bancroft, J.C., 2002, Event detection in prestack migration using matched filters, CREWES
Research Report, 14
Claerbout, 1992, Earth Soundings Analysis, Blackwell Scientific Publications.
Ziemer, R.E., and Tranter, W.H., 1988, Principles of Communications, John Wiley and Sons, pages 465-
468.

Matched Filterss

Uploaded by

Copyright:

Available Formats

Matched Filterss

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Matched Filterss

Uploaded by

Copyright:

Available Formats

Introduction to matched filters

Introduction to matched filters

1D model for matched filtering

Filter h(t) y(t) = sh(t) +nh(t)

FIG. 1: Basic filter

CREWES Research Report — Volume 14 (2002) 1

Example with a short wavelet

FIG. 2: Wavelet, only first 100 ms (50 points) used.

2 CREWES Research Report — Volume 14 (2002)

CREWES Research Report — Volume 14 (2002) 3

Example with a longer chirp wavelet

FIG. 5: Chirp wavelet.

FIG. 6: Noisy traces and chirp wavelet.

4 CREWES Research Report — Volume 14 (2002)

FIG. 7: Match-filtered result of the noisy chirps.

CREWES Research Report — Volume 14 (2002) 5

6 CREWES Research Report — Volume 14 (2002)

In migration, the diffraction stack algorithm cross-correlates a time and spatially

Converted-wave reflection amplitudes tend to zero at zero-offset and increase in

CREWES Research Report — Volume 14 (2002) 7

8 CREWES Research Report — Volume 14 (2002)

You might also like