Phase Unwrapping for Very Large Interferometric Data Sets

Phase unwrapping is one of the most challenging steps in synthetic aperture radar (SAR) interferometry processing. With the rapid advancement of SAR technologies, the interferometric data sets from newly launched satellites are becoming larger and larger. When the computing resources required for unwrapping an input data set exceed computer hardware capabilities, phase unwrapping becomes even more problematic. In this paper, a new method is proposed in order to address the problem of unwrapping large data sets. The proposed method separates the unwrapping procedure into two stages. First, an approximate L¹-norm phase unwrapping solution is efficiently obtained from a simplified minimum-cost flow network. Then, the blocks partitioned from the input data set are unwrapped by solving the corresponding independent network optimization problems directed by the approximate solution, either in parallel or in series. By then simply aligning the unwrapped blocks, a full-size unwrapped result is obtained. A significant advantage of the proposed method is that the globality of phase unwrapping solutions can be guaranteed. Using an interferogram with a size of 120 000 × 9274 pixels, the authors demonstrate that the proposed method is able to efficiently unwrap very large interferometric data sets using limited computing resources.