相位解缠作为SAR干涉测量数据处理中的一个关键步骤，受到越来越广泛的关注，出现了各种各样的算法。但现有的相位解缠算法仍无法解决高噪声问题，由此导致噪声区域的误差传递到其它区域，产生全程误差，从而影响相位解缠精度。针对这种情况，我们根据网络优化原理，提出了一种基于不规则网络下网络流算法的解缠方法，以干涉相干作为评价相位质量的标准，从含有大量噪声的干涉纹图中剔除低质量的相位，只对高质量相位进行处理，最终获取有用信息。该方法可以避免低质量区域的误差对高质量区域解缠的影响，保证高质量区域的相位解缠，从而获得较理想的解缠结果。

As the key step of InSAR data processing, many phase unwrapping algorithms were introduced. But these algorithms had difficulties with high noisy interferogram. This paper introduced a new method based on netowrk flow algorithm in irregular network, out of regard for high interferometric coherence and neglecting low coherence data area. First, a new phase data set with high quality data is created. Delaunay triangulation network can be derived according to the positions of the high quality phases in the image and the residue of each triangulation in the network is calculated. According to the positions of the residues in the irregular network the flows are computed by the network flow algorithm, we can integrate along the flows and derive the unwrapped phases. Finally, we compare the unwrapped phase image computed by network flow method with the image by traditional path finding method. From the comparison, we can conclude that the network flow method can increase the accuracy of the phase unwrapping and decrease the error occured noise.