Depth Information from Holographic Radar Scans

Holographic radar has several potentially important advantages over conventional pulsed radar for buried object location and identi¯cation. In particular the sensor can be lightweight, of low cost and provide images of the object with good resolution. This work reports that it can give some depth information also. Two simple experiments using the Rascan system operating at 5 discrete frequencies between 1.5 and 2.0 GHz are described. The ¯rst calibration" experiment used an aluminium plate buried in sand and inclined at a known angle to give a range of depths between 0 and 85 § 5mm over a length of 300 mm. The sensor was manually moved on a thin plate of glass placed over the sand above the plate. The scan down the plate showed characteristic bands in the backgound-corrected re°ected intensity | the zebra e®ect". The ¯ve re°ected amplitudes at each frequency are seen to vary di®erently with depth. These are in- terpreted qualitatively in terms of the simple theory involving interference between the re°ected wave and the incident wave at the sensor position. This theory does demonstrate the zebra e®ect and suggests that the amplitude variation with frequency is characteristic of the depth. The cyclic nature of the zebra stripes means that the more frequencies available, the better the depth discrimination. Quantitatively the true theory is complex, especially in the presence of the glass. However we suggest that the measurements can at least be used as calibration signals for a metal re°ector at a given depth. The second est" experiment involved nine US pennies buried in sand at known depths between 0 and 56mm with a lateral separation of about 50 mm. A background-corrected total re°ected intensity (the summed modulus of the re°ected amplitude over all frequencies) revealed the outline of each penny quite distinctly. For each penny the set of re°ected signal amplitudes at each frequency was determined at the position over the penny where the signal was maximized. This est" response for each penny gave a response over the ¯ve frequencies which was distinctive and is indeed characteristic of the depth. A precise simu- lation theory is not available but the best frequency response of each penny could be compared with the frequency response curves from the aluminium plate as a function of the known depth. A least squares ¯t was made between the best amplitudes of as a function of frequency for each penny, compared to the amplitude variation of each frequency as a function of known depth in the inclined aluminium plate experiment. With arbitrary amplitude scaling a least squares residual minimum as a function of depth was always obtained for any one penny. There were some problems. The amplitude scaling varied for each penny, presumably as its surface quality and orientation altered the intensity re°ected into the sensor. Also the signal quality deterio- rated with depth, and the frequency variation became less distinct at depths greater than 4 cm. However for the six pennies covering the depth range up to 3.4 cm the ¯tted depth agreed with the actual measured depth to within a standard deviation of 3 mm.