Ground Penetrating Radar Experimental Results
Please forgive the disjointed feel of this page. It was done to quickly record/document these results.
These are the images generated from actual, experimental data obtained from the Ground Penetrating Radar. Each image is an integration of 30 time domain signals.
The results present the image from unprocessed data and the best image generated using different basic signal processing techniques.
The techniques used were:
- ABS: Taking the absolute value of the input signal
- Normalizing: Normalizing the input signal for 1/r squared path loss
- Match filtering: Correlating the return signal with the input waveform (none currently shown)
- Ball/Target Correlation: Correlating the return signal with a simple model of the target
Please note that the theoretical maximum signal difference between the ball/target return is only 3 dB higher than the sand in this experiment. It should not come as a surprise that the unprocessed images are not exactly clear.
2 inch Ball Bearing Images
| Unprocessed data |
ABS, Normalized and Ball Correlated
(No, I don't see anything either...) |
|
|
2 inch Ball Bearing With Sand Background Subtracted Out
| Unprocessed data |
ABS
(That looks about right to me!) |
| Image missing
|
|
12 inch Pie Plate Images
| Unprocessed data |
ABS
(No noteable result here either) |
|
|
12 inch Pie Plate with Sand Background Subtracted Out
| Unprocessed data |
ABS, Normalized |
ABS, Normalized and Target Correlated
80 degree view |
|
|
|
Click here to see our GPR detect sand (a non-metalic substance).
Click here to see our GPR modeling results.
Click here to see the time domain data that generated these results.
Initial work by Bill Sunlin and Brian Hunt
Current model, implementation, exploitation and results by Donald Kerns
3D-Plots generated using Khoros by Khoral
dak 4/22/97