Imaginary numbers are just as real as logic is. Sure, there's no physical qualities to them (with the exception of neurons thinking about either one) but they still describe reality objectively.

Here's the thing with the concept of differing perceptions. For sight, photons of a fixed wavelength bunch off of surfaces, enter into our eyes and reach our optic nerves which are then processed to create the perception of color. These mechanisms are fixed across all humans. The wavelengths are constant, the method of interpreting this data is constant. So the only disconnect here is "what if the color I see is different than the color you see?" At this point science has come short of showing the connection between cognition and neurology, but the point remains the same: If everything in this line of causation is constant, why should we assume that the last part (the perception) is relative? "Don't add quantities unnecessarily" i.e. Occam's Razor. Sure it can't be "proven" as of this point in time, but what's more likely? That causation remains constant or that the last event in that chain somehow changes person to person? Furthermore, you have to prove the assertion that there's some mechanism in the brain that is allowing for these otherwise objective measures (wavelengths) to be perceived differently per person. "What you freely assert, I freely dismiss.

As for those scientific tests you mentioned, without seeing them my self (which I would like to if you could link me, sounds interesting) I'm sure they're something along the lines of "I see this as a bunny, but you see this as a rabbit" or "that looks pink to me and purple to him." At that point, that's just a matter of the state of the brain of the individual being different at some point during that chain of perception. No, I don't have the neuroscience behind that, but it's a rather reasonable assumption.

Also, to say that there's far more evidence "proving" that everything in reality is relative is faulty for two reasons: first, there's just as much work proving otherwise from philosophical writings and second, to "prove" anything assumes some objective constants which results can be measured against.