Faint, hidden contaminants in the point-spread functions (PSFs) of stars cause shifts to their measured positions. Wilson & Naylor (2017MNRAS.468.2517W) showed failing to account for these shifts can lead to a drastic decrease in the number of returned catalogue matches in crowded fields. Here we highlight the effect these perturbations have on cross-matching, for matches between Gaia DR2 and WISE stars in a crowded Galactic plane region. Applying the uncertainties as quoted to Gaussian-based astrometric uncertainty functions (AUFs) can lead, in dense Galactic fields, to only matching 55% of the counterparts. We describe the construction of empirical descriptions for AUFs, building on the cross-matching method of Wilson & Naylor (2018MNRAS.473.5570W), utilising the magnitudes of both catalogues to discriminate between true and false counterparts. We apply the improved cross-matching method to the Galactic plane |b|<10. We provide the most likely counterpart matches and their respective probabilities. We also analyse several cases to verify the robustness of the results, highlighting some important caveats and considerations. Finally, we discuss the effect PSF resolution has by comparing the intra- catalogue nearest neighbour separation distributions of a sample of likely contaminated WISE objects and their corresponding Spitzer counterpart. We show that some WISE contaminants are resolved in Spitzer, with smaller intra-catalogue separations. We have highlighted the effect contaminant stars have on WISE, but it is important for all photometric catalogues, playing an important role in the next generation of surveys, such as LSST.