A Cluster-Analysis-Based Noise-Robust Phase-Unwrapping Algorithm for Multibaseline Interferograms

Two-dimensional phase unwrapping (PU) is a key step of synthetic aperture radar interferometry (InSAR). Moreover, the conventional single-baseline PU method is restricted to the phase continuity assumption, so it cannot work correctly in the case that phase jumps between adjacent pixels are larger than π. To effectively solve this problem, multibaseline PU is put forward. The performance of conventional multibaseline PU methods is directly related to the noise level. In order to improve noise robustness, a cluster analysis (CA) based noise-robust PU algorithm for multibaseline interferograms (CANOPUS) is proposed in this paper, which is the extension and improvement of the CA-based efficient multibaseline PU algorithm proposed by H. Yu. For the sake of overcoming the disadvantages of the CA method, the dimension of the recognizable mathematical pattern is expanded. Under this condition, due to the density discrimination in spatial space, different clusters are able to be distinguished by the density-based clustering algorithm, and clusters are regarded as a set of density-connected patterns. Compared with the conventional CA method, the significant advantage of the new algorithm is that it improves noise robustness. What is more, the proposed algorithm runs in linear time. From the experiment results, it can be seen that the proposed method may be effectively applied to multibaseline InSAR data sets.