Abstract:

In the interpolationbased methods for subpixel registration of interferometric synthetic aperture radar (InSAR) images, but the registration accuracy is restricted by the interpolation unit, and the computational burden is heavy when high accuracy is demanded. The proposed method implements the analytical optimization of cross correlation to overcome the limitation aforementioned. Firstly, a novel analytic cost function is derived by combining the interpolation operation with the crosscorrelation searching process. The offset is searched in the continuous domains, which suggests a more accurate registration since the estimation accuracy does not depend on the interpolation unit. Then, by incorporating the gradient information of the cost function, the quasiNewton method is employed to optimize the cost function. The subpixel offset associated with the maximum of the cost function can be quickly obtained with low computational complexity. The experimental results show that the proposed method has higher registration accuracy and superior performance in terms of registration accuracy and computational complexity when compared with several conventional methods.