Proposal of bilinear surface compensation of distortion in least-squares phase unwrapping

The least squares phase unwrapping (LSPU) is known as one of the useful phase unwrapping (PU) methods to obtain digital elevation map (DEM) from interferometric synthetic aperture radar (InSAR) phase image. However, the LSPU process looses the constant component and the first derivative component contained in the InSAR data. Though the constant component loss is trivial, the loss of the first derivative values possibly causes a serious global distortion in the resulting DEM because its effect appears not as a plane slant but as a bilinear curved surface composed of low frequency components. In this paper, we present that the bilinear-surface distortion is often a dominant cause of DEM degradation. We present an experimental result for an artificially generated phase image free of singular points to clarify the LSPU processing effect. Then we propose the method to estimate a correct bilinear surface to compensate the distorted DEM.