A 3-D Minimum Cost Flow Phase Unwrapping Algorithm Based on Closure Phase for InSAR
1.Required third party package
You should have Triangle (https://www.cs.cmu.edu/~quake/triangle.html) installed to perform our algorithm. This package is used to generate Triangulated irregular network (TIN) to connect discrete pixels.
2.Data prepation
Please read matlab/InputParameters.m for more details. Currently, the updated code only works for discrete dataset.
3.Optimization
Currently, we are using a built-in function 'linprog' in Matlab to solve our equation, which however is time and memory consuming and sometimes may fail. Currently I am planning to write this part of code by myself, so if you know how to solve the equation from mathematic point of view, please let me know ([email protected]). In addition, if you are using Matlab version after 2012a (the version which we used to develop our code), you may need to add the 'optim' dir to your path as the 'linprog' function has updated and may casue some problems in the later Matlab version.
4.Test_data
We recently add a sample data for quick test, which contains three interferograms that form one phase closure (from real Sentinel-1 data). You can use the 'scatter' function to plot the wrapped or unwrapped data, for example:'scatter(coor(:,1),coor(:,2),3,'filled','cdata',uph(:,1));'.
Recent changes:
licols.m: Fix the bug that can't extract linearly independent set of columns when the matrix has only one column (this might be related to the version of matlab you use).
If you have used our code, please cite our paper:
F. Liu and B. Pan, "A New 3-D Minimum Cost Flow Phase Unwrapping Algorithm Based on Closure Phase," in IEEE Transactions on Geoscience and Remote Sensing, vol. 58, no. 3, pp. 1857-1867, March 2020, doi: 10.1109/TGRS.2019.2949926.