You only need to calculate the misorientation: Ga*Gm-1.  That's all (as written in many books, articles, internet). 

The "only" problem is that many people already spent years to find the OR between austenite and martensite, however there is not only one. S. Zaefferer from your institute (maynbe even your group) knows this very well :-). He wrote and article with Dr. Kato a few years ago about some kind of continuous change of the OR around a single precipitate within a matrix. You can imagine what happens when all the matrix is gone and you only have the martensites which produce very blurred patterns (because of inherent lattice defects. Thus, the OR is changing within martensite (or at least it does would not surprise me). 

Another major problem is the limited precision and accuracy of EBSD patterns using fast indexing by Hough transform. I am not very optimistic that the clearly higher precision of pattern matching will change much for this material but I don't want entirely exclude it.

There are several tools which already try to reconstruct austenite from magnetite/ferrite precipitates (Cyril Cayron with his program ARPGE (now in Swiss), a group in Metz, another in Birmingham, and also one guy in Japan). All have more or less the same problem: the recognition of the variants to a specific austenite grain since one variable is the OR itself. In steel I havn't seen K-S or N-W., I mean the exact pole distribution The ORs are always in between. For other systems the chemical segregation obviously causes a better adaptation of fcc an bcc planes so that even recently in Co-Al alloys we measured it. At least it was very close to K-S. From a single variant I wouldn't try it. I have seen in the same microstructure also Greninger-Troyano (95%) and Pitsch (5%). Tey are quite different so that any doubt should be excluded. But there is one opinion that in steel you will find anyway "everything" so that e.g. Cyril Cayron recently published a paper where he discussed again this idea first (?) introduced by Bunge in (2002). in Taking the orientation variation along a precipitate, the limited accuracy and the deviation from the well-known models into account it seems to be that you have a very challenging topic. Good luck!