Thread Subject: Alternative to BWlabel?

Hi, not sure if this will yield useful results since there
is already a function (albeit one that I do not have). I
have a logical matrix created from a bandpassed image. This
matrix will have several regions of connected 1's. Is there
any way (other than BWlabel) to identify these areas
independantly? The end goal would be to seperate them into
individual matrixes.
Ex:
[ 0 0 0 0 0 0
  0 1 1 0 0 0
  0 1 1 0 0 0
  0 0 0 0 1 1
  0 0 0 1 1 1 ]
turns into:

[ 0 0 0 0 0 0
  0 1 1 0 0 0
  0 1 1 0 0 0
  0 0 0 0 0 0
  0 0 0 0 0 0 ]
and
[ 0 0 0 0 0 0
  0 0 0 0 0 0
  0 0 0 0 0 0
  0 0 0 0 1 1
  0 0 0 1 1 1 ]

Right now I am dealing with 128 x 128, and the objects will
be irregular shapes. Any ideas are appreciated!

"Jon Schneider" <j@yahoo.com> wrote in message
<fj77gu$1q2$1@fred.mathworks.com>...
> Hi, not sure if this will yield useful results since there
> is already a function (albeit one that I do not have). I
> have a logical matrix created from a bandpassed image. This
> matrix will have several regions of connected 1's. Is there
> any way (other than BWlabel) to identify these areas
> independantly? The end goal would be to seperate them into
> individual matrixes.
> Ex:
> [ 0 0 0 0 0 0
> 0 1 1 0 0 0
> 0 1 1 0 0 0
> 0 0 0 0 1 1
> 0 0 0 1 1 1 ]
> turns into:
>
> [ 0 0 0 0 0 0
> 0 1 1 0 0 0
> 0 1 1 0 0 0
> 0 0 0 0 0 0
> 0 0 0 0 0 0 ]
> and
> [ 0 0 0 0 0 0
> 0 0 0 0 0 0
> 0 0 0 0 0 0
> 0 0 0 0 1 1
> 0 0 0 1 1 1 ]
>
> Right now I am dealing with 128 x 128, and the objects will
> be irregular shapes. Any ideas are appreciated!


I bought the image processing toolbox a couple of weeks ago
but I had a quick look on the web just now and found this
'alternative'.

http://www.petercorke.com/Machine%20Vision%20Toolbox.html

There's a nice PDF describing the toolbox available here

http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=1577021

Perhaps iblobs and ilabel will do what you need?

Ken

"Ken Campbell" <campbeks@gmail.com> wrote in message
<fj7mds$sis$1@fred.mathworks.com>...
> "Jon Schneider" <j@yahoo.com> wrote in message
> <fj77gu$1q2$1@fred.mathworks.com>...
> > Hi, not sure if this will yield useful results since there
> > is already a function (albeit one that I do not have). I
> > have a logical matrix created from a bandpassed image. This
> > matrix will have several regions of connected 1's. Is there
> > any way (other than BWlabel) to identify these areas
> > independantly? The end goal would be to seperate them into
> > individual matrixes.
> > Ex:
> > [ 0 0 0 0 0 0
> > 0 1 1 0 0 0
> > 0 1 1 0 0 0
> > 0 0 0 0 1 1
> > 0 0 0 1 1 1 ]
> > turns into:
> >
> > [ 0 0 0 0 0 0
> > 0 1 1 0 0 0
> > 0 1 1 0 0 0
> > 0 0 0 0 0 0
> > 0 0 0 0 0 0 ]
> > and
> > [ 0 0 0 0 0 0
> > 0 0 0 0 0 0
> > 0 0 0 0 0 0
> > 0 0 0 0 1 1
> > 0 0 0 1 1 1 ]
> >
> > Right now I am dealing with 128 x 128, and the objects will
> > be irregular shapes. Any ideas are appreciated!
>
>
> I bought the image processing toolbox a couple of weeks ago
> but I had a quick look on the web just now and found this
> 'alternative'.
>
> http://www.petercorke.com/Machine%20Vision%20Toolbox.html
>
> There's a nice PDF describing the toolbox available here
>
>
http://ieeexplore.ieee.org/xpl/freeabs_all.jsp?arnumber=1577021
>
> Perhaps iblobs and ilabel will do what you need?
>
> Ken
>

Thanks! That looks like exactly what I need.
This was my eventual solution:
%index is the index to the binary. This program sorts each
%connected region to nindex (then just delete the extra
%zeros that are put in there).

n=1;
m=size(index);
while ~isempty(index)
    newindex=index(1);
    a=[1 2];
    while a(2)>a(1)
        for i=1:size(index)
            if sum(ismember(newindex,index(i)-1)) ||
sum(ismember(newindex,index(i)+1)) ||
sum(ismember(newindex,index(i)-128)) ||
sum(ismember(newindex,index(i)+128))
                newindex=[newindex;index(i)];
            end
        end
        index=index(end:-1:1);
        newindex=unique(newindex);
        a(1)=a(2);
        pop=size(newindex);
        a(2)=pop(1);
    end
nindex(:,n)=zeros(m(1),1);
nindex(1:size(newindex),n)=newindex;
index=setxor(index,newindex);
n=n+1;
end

If anyone has any comments for improvement (efficiency) it
would greatly help out a beginner.

Thanks again.