Releases: kanglcn/moraine
v0.9.0
v0.8.5
v0.8.3
Move two deep learning models out of the package to prevent too big pypi package
update cli.co use image_pairs in memory rather than tnet file;
fix plot bugs in functions for general read zarr file, and change seq_intf from (i, i-1) to (i,i+1);
add multilook and intf to api.co;
finish gamma phase wrapping API and CLI;
move read/write gamma file to a seperate module from cli.load
v0.8.2
bug fixed for cli.emperical_co_pc
Bugs Squashed
- Data format scomplex is not supported. (#20)
-
Description of the problem
-
The package doesn't support scomplex data, so if the preprocessed rslc data is saved as scomplex, you are supposed to convert it to fcomplex and then process it with this package.
Minimal Complete Verifiable Example
No response
Full error message
No response
System information
numpy version 1.22.4Are you willing to help fix this bug?
No
v0.8.1
implement parallel zarr io
modify emperical_co_pc and necessary utils
for independent ras chunkwise processing
add emperical_co_emi_temp_coh_pc to
- only processing a small batch in one chunk to prevent
holding the coherence matrix for all chunk which may exceed
memory limit; - prevent writing coherence matrix which may exceed disk space limit.
v0.8.0
- add cpu version of all functions and make use cpu as default
- add default args to dask cluster and allow users to configure that
- only calculate low tri of coherence matrix and copy its conj to up tri
- more flexiable plot functions
- carefully deal with nan values for all functions
- Update
pl.temp_cohwith elementwise kernel