Chooses feature points suitable to compute optical flow image correspondences.
calibration/cg_optical_flow_features dataset_parameters.json [refgrid.json] [out_fpoints_dir/] [dataset_group]
Displays a GUI where parameters can be adjusted, and the chosen feature points are shown. The parameters from cv::goodFeaturesToTrack can be adjusted. Also, to well-distribute features in the image, it can first subdivide the image into n rectangles, select features on each one, and keep all the best features.
If a references grid refgrid.json is given, chooses the feature points on all the views of the references grid. Hitting Enter generates a feature points file for each reference view in the directory out_fpoints_dir/. If the references grid is not give, one view can be freely selected.
The features will have unique names like feat_RFFF, where R is the number of the reference view, and FFF a number of the feature.
The optical flow should then be computed using calibration/cg_optical_flow_cors, once for each of the feature points files.