I can't speak to suitability for your application. I don't know the material of your axle, or of any heat, forging, shot peening or other processes which give it it's current properties, nor do I know what loads your axle might sustain after your proposed modification.
The referenced paper gives no direct information of use. It does suggest that for a thermal camera, good estimates of emissivity can be obtained, but you would need two apply the methods described in the paper to the two different bands used by the UniT unit and compare their differences.
Most of the unknowns in your question can only be answered by you. What heats do you you need. How do you accurately do you need to know them. Is uniformity of temperature needed, or do you needed a controlled pattern.
I can only give a general answer. I have a UniT thermal viewer and it gives results I believe to be within about 10-20% for temperatures up to a couple hundred degrees. And I have several spot thermometers similar to the UniT (possibly off the same assembly line) which provide results that are comparable. I am not saying that these devices have errors of 10-20%, I am merely saying that I have no sound evidence that they are better than that.