The point of my thesis is that it has not been done by anyone else before. In fact, I had to do the mathematics because previous work in the area required too much training data to be effective for mapping; it is simply not viable to require dozens of training passes through an area before I can begin to detect anomalies.
My algorithm works because it uses assumptions from the physics of the problem -- the assumption of Poisson-distributed counts.
Well, I can't argue with the specifics of your problem since I don't know about it (but the general point that this doesn't justify the current maths system for 99% of people).
However, at some point in time, computers will be taking over all human abilities (if you believe that AI is inevitbale). At what point will that occur? It's my strong suspicion that we have already reached that point and any evidence to the contrary is simply that people aren't trying automated methods (I believe that machine learning can solve any problem that humans can solve, and furthermore, this wouldn't require more computing power than we already have).
My algorithm works because it uses assumptions from the physics of the problem -- the assumption of Poisson-distributed counts.