Recent proposals suggest that distributed single photons serving as a ‘non-local oscillator’ can outperform coherent states as a phase reference for long-baseline interferometric imaging of weak sources. Such non-local quantum states distributed between telescopes can, in-principle, surpass the limitations of conventional, visible wavelength, interferometric-based astronomical imaging approaches for very-long baselines such as: signal-to-noise, shot noise, signal loss, and faintness of the imaged objects. Here we demonstrate in a table-top experiment, interference between a non-local oscillator generated by equal-path splitting of an idler photon from a pulsed, separable, parametric down conversion process and a spectrally single-mode, quasi-thermal source. We compare the single-photon, non-local oscillator to a more conventional local oscillator with uncertain photon number for a few different source distributions: Gaussian, 1 mm separated double slits, and 2 mm separated double slits. Both methods were able to return a source reconstruction or the source’s autocorrelation. We find that the non-local oscillator has a higher signal-to-noise ratio per detected coincidence compared to a classical local oscillator; confirming the reduction in shot noise by using a non-local oscillator.