There is one quick way to do it, however it is based on the assumption that the initial array will always have only 3 distinct values:
minV=min(H); % find min value in H
H(H==minV)=0; % set all min values to zero
maxV=max(H); % find max value in H
H(H==maxV)=0; % set all max values to zero
At this time you should have intermediate values somewhere in H. To separate them in the new matrix, you do the following:
Intermediate=H(H~=0);
This will give you the matrix with non-zero values, that is max and min values are eliminated.
If you want to get the actual value from the new H matrix you can do the following:
ind = find(H~=0, 1, 'first');
finds the address of the first non-zero value;
yourValue=H(ind)
gives the first non-zero value that is assumed to be the intermediate value between min and max.
