Metal detector image preprocessing

Surprisingly, the metal detector, which is the most common detector, considered as an imaging device, can also provide very useful information on the shape of metallic pieces included in mines. Trials have been performed in the RMA [16] by placing the Vallon metal detector on a gantry. Thanks to the collaboration of Vallon, the metal detector is provided with a digital output. The obtained images were processed as follows: after noise reduction using a regularized Wiener filter, the images were deconvolved using respectively an experimental point spread function (PSF), recorded using a small metallic ball, and a simplified theoretical model of the PSF, calculated from the detector head shape. From left to right, fig.(18) shows successively the original image, the experimental PSF and the result after deconvolution. Fig.(19) shows the same but using the theoretical PSF.
A practical example, using the exercise mine PRB409 which has a diameter of 7.5 cm, is shown on fig.(20). The image on the left is the original one. The central image is a RX photography of this mine where one sees clearly a small horizontal metallic cylinder on the right side. The image on the left shows the deconvolved image where the metallic cylinder clearly appears.

  

Figure 18: Metal detector - deconvolution with experimental PSF

  

Figure 19: Metal detector: deconvolution with theoretical PSF

  

Figure 20: Metal detector -- PRB mine deconvolution

Unfortunately, the PSF of a metal detector is a function of the depth and the image formation process is non linear. However, it has been shown by the RMA team that it is possible to derive the depth of a buried object from the original data and thus to derive the corresponding PSF to allow a correct deconvolution. This subject is always under investigation. This interesting consideration shows that the metal detector, known as a cheap mine detection system, is still a promising device.