SAR image registration in the presence of rotation and translation : a constrained least squares approach

This letter proposes a coregistration algorithm to compensate for possible inaccuracy of trajectory sensor during the SAR image acquisition process. Such a misalignment can be modeled as a pure displacement in range and azimuth directions and a rotation effect due to different angle of sight. The approach is formalized as a Constrained Least Squares (CLS) optimization problem enforcing a constraint of absence of a zooming effect between the two SAR images. Moreover, system equations can optionally be weighted according to local properties between the extracted patches within the quoted couple. Interestingly, the solution can be obtained in closed-form, therefore with a low computational cost. The results of the tests conducted on the 9.6GHz Gotcha SAR data demonstrate the capability of the strategy to proper register the imagery.