Also, K. pneumoniae is intrinsically resistant to ampicillin. You can use a selective medium for gram negative bacteria (like McConkey) with the antibiotic. This will probably reduce considerably the number of recovered bacteria from your filter.
I don't think there is any direct extraction method to isolate a single type of microbe out of the microbial population. But you can screen them using different microbiological assays specific for K.Pneumoniae which is a bit time consuming but if you have MALDI-MS you can easily identify the microbial strain.
Also, K. pneumoniae is intrinsically resistant to ampicillin. You can use a selective medium for gram negative bacteria (like McConkey) with the antibiotic. This will probably reduce considerably the number of recovered bacteria from your filter.
To add to Sitara Nasar 's answer, if you want to go the MALDI-TOF route for identification, you will need to isolate single colonies with selective media and/or drawing the sample across the plate in such a way as to allow single colonies to form. If you were to mix colonies, it could confound your MALDI results. Additionally, be careful what kind MALDI analysis you want to approach. Closely related species may be difficult to differentiate with something like Biotyper (though other methods may also prove difficult).
For just isolation, Fernando's answer is excellent.