2-DIMENSIONAL INTERPOLATION IN THE WAVE-NUMBER DOMAIN FOR RADAR IMAGE-RECONSTRUCTION

In the two-dimensional radar image reconstruction using diffraction to mography, missing data become a significant problem if the observation direction and the scan range are restricted. A method is proposed in which these missing data are interpolated in the wavenumber domain. As an object image, a collection of the point targets is assumed. By mea ns of a method which minimizes repetitively the square residual of the observed values and the reconstructed image in the wavenumber space, it is shown by simulation that a good image reconstruction is possible even if the missing rate is as high as 50 percent. As a method to pro vide the initial value for the repetitive method, an interpolation met hod using the autocorrelation function is proposed. It is shown that t his scheme works well in a stable manner even for a high missing rate and a poor SN ratio. Further, an objective discrimination criterion is proposed for estimation of the number of targets. It is shown that th e true target can easily be discriminated if the SN is good and that a stable estimation is still possible even if the SN is poor and the ex traction of a weak image is difficult.