Though I've never done it I've heard that impregnating silica with silver nitrate can help separate out E/Z isomers due to pi-Ag interactions being different on the solid phase.  I would assume dissolve AgNO3 in a polar organic solvent, add silica and then remove the solvent.  Best of luck

E and Z forms of the benzylidine group will rapidly equilibrate, as Allan points out

The barrier to inversion of the acyl hydrazone is structure-specific (see link), and some species will indeed interconvert at room temperature.  I'd expect the stable ones to be separable by HPLC, although preparative chromatography might be challenging.  Selective crystallization is possible, but entirely unpredictable.

http://pubs.acs.org/doi/abs/10.1021/jacs.5b09519

Silica gel impregnated with silver salts is sometimes being used for separating E/Z isomers of olefines (C=C bond isomers); here we have C=N bonds, and their coordination to Ag could be different. Quite often C=N geometrical isomers are separable without the need of specially treated silica (or any other stationary phase, especially if you use modern HPLC). However, due to the mentioned above isomerization, this separation might not be feasible.

By the way, I would not call C=N (Z/E)-isomers interconversion "racemization" since this term is usually reserved for the enantiomers of chiral compounds.

May be you can useful from the following article

http://nopr.niscair.res.in/bitstream/123456789/16218/1/IJCB%2038B%282%29%20192-196.pdf

I think the main point you first need to consider is... do these imine "cis/trans" isomers even exist as stable, single diastereomers? I tend to agree with the above that, aside from cooling or using other trapping agents, you will always have a mixture of these compounds due to rapid isomerization. I would be interested to hear (if you do have confirmation of 4, separate isomers of these imine compounds) how that is being determined (HPLC, mp, etc.). Cheers.