Um, it's easy actually.
>"just because cameras dont help solve crimes, doesnt mean they dont help prevent crimes. its very hard to find evidence of a crime not being commited."
No it isn't, it's simple. Take two very similar areas with similar demographic profiles and crime rates. Install cameras in one and not the other. See what happens to the crime rates. If one of them falls, that's evidence of crimes not being committed.
See, thing is, you don't need to show /which particular/ crime was not committed; just that some - any - that would otherwise have been, were not. Or, as the case may be, that there is no particular subsequent difference between the two crime rates.
As has been mentioned elsewhere, the one thing they *do* do is help solve crimes after the fact - although as also mentioned elsewhere, that's little consolation to the victims, it isn't prevention, and it takes a thousand cameras for every one crime that is thereby solved.