Ice drift retrieval algorithm based on combination of normilized cross-correlation and phase correlation
Programmed by: Denis Demchev, Anders Hildeman, Eduard Kazakov, Anton Volkov
To start working with the package we recommend to create conda environment containing all required Python packages first:
$ conda create --name YOUR/ENVIRONMENT/NAME --file /PATH/TO/requirements.txt
and then activate it:
$ conda activate YOUR/ENVIRONMENT/NAME
- The main script: cc_bm_parallel_pyr_dev.py
cmd arguments: "filename1 filename2 block_size search_area grid_step"
We recommend to use the image patch size of 16-48 pixels that corresponds to 640-1920 ground meters. The searching zone of < 100 km per day is seems to be optimal to handle typical sea ice drift.
Example of usage in IPython:
run cc_bm_parallel_pyr_dev.py clip_HH_S1B_EW_GRDM_1SDH_20200301T083237_20200301T083346_020496_026D68_5471_adjusted.tif clip_HH_S1B_EW_GRDM_1SDH_20200302T073529_20200302T073629_020510_026DD5_27F9_adjusted.tif 64 4 30
1.1 Config file
cc_config.py
the file contain all algorithm parameters
1.2 The number of CPU's for parallel computing is defined in line 1274
1.3 cc_calc_drift.py
Calculate ice drift
1.4 cc_calc_defo.py
Calculate ice deformation
1.5 cc_calc_drift_filter.py
Erronemous ice drift vectors filtrering by homogenity criteria
1.6 The output of the algorithm is saved in "res_DATE_TIME"
You should get something like this:
