I have been trying to think about how to do this and I can't come up with anything. Imagine I have a data set of 220x220 rows and columns, below is a smaller version of what it looks like:

0000000000000000000000000000000000000000000000
0000000000000000000000000000000000000000000000
0000000002222222200000000000222222222200000000
0000000021111111220000000000221111111112000000
0000000211000001222222222222100000000012200000
0000002110000001111111111111100000000012000000
0000002110000012222222222222000000000120000000
0000000211111220000000000000211111111120000000
0000000002222220000000000000022222222200000000
0000000000000000000000000000000000000000000000
0000000000000000000000000000000000000000000000

where each of the 1's represent a sequence of numbers between 0 and 100, while everything outside this figure is 0. The sequence of numbers is large depending on where you are in the figure, but usually ranges from 10-15 numbers. My problem is the only numbers I want here are the 2's, which represent the boundary of the figure, I don't want the two little holes inside with the zeros. I have no problem getting the boundary of the figure, I just search through the elements until I come to a non-zero one and I check if any of its nearest neighbors are zero and if so then it represents a boundary piece.

The problem is that this method also gives me the inner boundary as well, where there are two holes cut out with zero elements. Can somebody think of a way just to get the 2's and nothing else. The numbers that aren't zero range between 1 and 100 and have no particular pattern to them.
Thanks if anybody can help.

Jeff