How to delete consecutive values in a vector and to save just the biggest one?

I have a vector like: a=[1 2 3 7 10 11 12]. The idea that i want to implement is: if I have consecutive numbers (in this situation) 1 2 3 and 10 11 12 to save just the biggest one and still to not delete the "7". So it should become: a=[3 7 12].

My code (it doesnt work properly and gives me error also):

a=[1 2 3 7 10 11 12]
k=numel(a);
for n=2:k
if a(n)==(a(n-1)+1)
   a(n)=[];
    end
end

Answer 1

[Stephen23]

If the sequences are integer and increasing only:

>> a = [1,2,3,7,10,11,12];
>> x = [diff(a)~=1,true];
>> a(x)
ans =
     3     7    12

[Miro Mitev]

Thanks :)

[Brian]

If anyone comes across this and is having issues ("horzcat" error) using vertically orientied data as I was, just transpose the array inside of the diff function. For example:

>> a = [1;2;3;7;10;11;12]
>> x = [diff(a')~=1,true];
>> a(x)
ans = 
     3
     7
    12

[Stephen23]

@Brian: it is simpler and more efficient to just concatenate along the first dimension:

>> x = [diff(a)~=1;true];
>> a(x)
ans =
     3
     7
    12