空间网络对时序InSAR相位解缠的影响——以Delaunay与Dijkstra网络为例EI北大核心CSCD

在时间序列InSAR相位解缠的过程中,相干目标需事先构建空间网络之后再进行模糊度解算。Delaunay三角网是当前时序相位解缠的主流构网方法,但其网络形态易包含高相位梯度的边缘,导致违背相位连续性假设。考虑到目前很少有关于空间网络对解缠影响的研究及相位解缠对InSAR技术测量精度的主导地位,本文在量化分析Delaunay网络对解缠影响的基础上,提出引入图论中的Dijkstra最短路径算法优化Delaunay网络中所有边的相位梯度,进而改善时序相位解缠的精度。本文采用模拟和真实数据对基于Delaunay网络和基于优化网络的相位解缠进行了对比验证。结果表明,本文提出的构网方法能够更好地满足相位连续性假设,减少约33%由解缠误差所导致的不闭合三角环数。较传统研究聚焦解缠方法和目标函数的改进而言,本文研究揭示了空间网络的改善对时间序列相位解缠的重要性。

Effects of spatial network on time series InSAR phase unwrapping:take the Delaunay and Dijkstra networks for example

In time series InSAR phase unwrapping,coherent pixels are required to be fully connected by a spatial network and then ambiguity can be estimated.During the network generation,Delaunay triangulation has been an invariable choice for InSAR community,but its network configuration easily contains the edge of high phase gradient,which leads to the violation of phase continuity assumption.Given that whether the geometry of Delaunay is suitable for phase unwrapping or not is rarely discussed,in this paper we quantitatively discussed the performance of Delaunay triangulation in phase unwrapping and we also proposed a new network through introducing the graph theory to improve the unwrapping accuracy.Experimental results validate that the proposed method can better satisfy the phase continuity assumption,and is superior to the Delaunay network for a reduction of~33%unclosed interferogram triangle loops.Compared to the previous researches focused on unwrapping methods and the improvement of objective function,this study reveals the importance of the improvement of spatial network to the time series phase unwrapping.