IFSAR correlation improvement through local slope correction

The author has developed a robust algorithm to help compensate for the phenomena of local slope decorrelation in SAR interferometric pairs. A standard maximum-likelihood algorithm assumes that the underlying terrain elevation within the computation window is constant. An improved maximum-likelihood correlation measure makes the assumption that the underlying terrain elevation within the moving window box is not constant, but is instead planar. Estimating the planar phase slopes is equivalent to measuring the instantaneous frequency of the interferogram in the moving window box of interest. The maximum-likelihood estimate of the planar phase is equivalent to finding the location of the peak of the Fourier transform of the interferogram data within the local computation window. Once estimated, the terrain can be "flattened" and the correlation computed on the "flattened" terrain, hence compensating for local slope decorrelation. Wrapped phase estimates are also significantly improved. This algorithm has also been extended to allow "space-variant aperture trimming" which offers additional improvement in the correlation estimates at the expense of space-variant resolution.