Multichannel fusion of subsurface radar images at different resolutions

The authors deal with the fusion of multichannel and multiresolution subsurface radar images collected by different frequency channels. Since each radar frequency is most suited for the investigation of a specific range of depths, an accurate analysis of the subsurface over an extended depth interval requires images at different channels. By exploiting the physical model of EM pulse propagation in a stratified medium, three multichannel Wiener filters are derived for estimation of the subsurface impulse response from a set of radar surveys collected at different frequencies. By properly combining the echoes received at the different channels, the filters yield the high resolution of the high-frequency channels for the shallow interfaces and the good SNR of the low-frequency channels for the deep interfaces. A complete theoretical as well as simulated performance analysis is presented for the proposed filters. Moreover, the fusion techniques are applied to real data collected at two sites: a homogeneous subsurface with a number of embedded, known, finite objects and a simple stratified natural terrain. The former is used to assess the proper operation of the multichannel deconvolution, while the latter is used to show that the desired optimal performance is achieved for the interfaces at different depths.