| 摘要 |
A method for monitoring terrain and man-made feature displacements comprising pre-processing sets of complex SAR images acquired in a first and in a subsequent campaign by means of a GBSAR instrument to obtain an incoherent mean SAR image for each campaign;
selecting, in each of the incoherent mean SAR images, pixels which are representative of targets respectively in each of said sets of SAR image data, the said targets being selected from artificial corner reflectors, if they are deployed in the area of interest, and natural reflectors in the area of interest;
performing a global image matching over the selected pixels to obtain a set of pairs of global shifts expressed in pixels and selecting a plurality of quality-scored pairs of global shifts;
estimating GBSAR repositioning effects by obtaining a mask comprising stable-area pixels corresponding to stable areas in the area of interest, computing a subset of global shifts falling in the mask, estimating a co-registration transformation on the grounds of the subset of global shifts, and producing a set of co-registration parameters modelling the GBSAR repositioning effects;
estimating pairs of displacements for each campaign by subtracting the GBSAR repositioning effects modelled by the co-registration parameters from the set of pairs of global shifts, thereby obtaining a plurality of pairs of displacements shifts expressed in pixels which, converting the displacement shifts expressed in pixels into displacements expressed in a displacement unit, and obtaining a set of pairs of estimated displacements defined in an object space by geocoding involving an image-to-object transformation. |
