First of all, I’d advise you, if you did not yet, to try reassign those secondary structure elements with DSSP. PyMol at least is using its own algorithm to favor aesthetics over accuracy (in overall it is really good but you never know…).
If those structured regions are at the interface, you can indeed tell HADDOCK to define dihedral restraints for 1) all backbone, 2) only alpha-helices or 3) only alpha-helices and beta-sheets. This can be done in the Guru interface, in the
Energy constants for hydrogen bond restraints sub-section of the
Restraints energy constants section.
Hope this will help,