Taking data from two different sources with completely different methods of obtaining their data is inherently unreliable, especially when one has a bias towards undercounting and the other towards overcounting and moreover they don't compare incidents of equivalent severity. In short, the comparison you make relies upon junk statistics of the sort only a university social sciences professor could love.
I'm curious as to why you think that the Bureau of Justice Statistics is biased towards undercounting (I'm not saying that it isn't, just wondering why you think it is).
Even if the data has some inaccuracies (as does most of this sort), do you really believe that they're extreme enough to negate the considerable gap between the two bars on the chart, or to provide credence to Bedstuy's assertion that almost no firearms are used to prevent crime?