Yes, in general sub-pixel (or sub-sample) correlation is possible. See this paper by Avouac et al.: ftp://white-dome.gps.caltech.edu/pub/avouac/KashmirEQ/Avouac-EPSL-2006.pdf
From Unwrapped SAR Interferogram Line of Sight (LOS) component & from the sub-pixel correlation of optical imagery, NS and EW true horizontal components is derived. Is it possible to get the true vertical component from these three derived components?