1) The function reorder_classes should provide an easy template to compute these differences. Copy it , place following the existing reorder_classes function and rename something like pair_wise_differences.
2) A new conditional read from the parameter file will be needed to enable. Something like "Cls_save_pair_wise_diffs"
3) Modify the current loop structure in that function to return a difference rather than a cross-correlation coefficient.
4) Write out the results to disk. Include the normal prefix _pwdiffscls#_cls#
5) It should be fine to average eve/odd for this prior to diff
It is probably useful to do this in the cache directory, or to make a new folder class_avg_pair_wise_diffs (or @something.) @alisterburt
In BH_average.m there is a block that is currently not used, that does a pairwise CCC comparison of volumes to reorder even/odd half sets
1499 - if strcmpi(STAGEofALIGNMENT, 'Cluster') && (flgClassify ~= -1) [classListOut, geometry] = reorder_classes(filteredClass(:,1),filteredClass(:,2),maxClasses, geometry); . . .
1) The function reorder_classes should provide an easy template to compute these differences. Copy it , place following the existing reorder_classes function and rename something like pair_wise_differences.
2) A new conditional read from the parameter file will be needed to enable. Something like "Cls_save_pair_wise_diffs"
3) Modify the current loop structure in that function to return a difference rather than a cross-correlation coefficient.
4) Write out the results to disk. Include the normal prefix _pwdiffscls#_cls#
5) It should be fine to average eve/odd for this prior to diff
It is probably useful to do this in the cache directory, or to make a new folder class_avg_pair_wise_diffs (or @something.) @alisterburt