How do I find the boundaries of a value in a matrix?

Hi, I have a matrix as follows:

I =

     1     1     1     1     8     1     2
     1     1     8     8     8     2     1
     1     8     8     1     2     1     1
     1     1     8     2     1     1     1
     2     2     2     1     1     1     1
     2     2     2     1     1     1     1

I want a matrix which will be output of boundary value of 8 i.e. (1,5),(2,3),(2,4),(2.5),(3,2),(3.3),(4.3)

Is there any way to he find these location easily?

Answer 1

[Sean de Wolski]

[r c] = find(bwperim(I==8));

[Mohammad Golam Kibria]

sorry, this also not giving me proper output.

[Walter Roberson]

That's because you keep changing the stated requirements :(

[Ivan van der Kroon]

This method and Andrei's still gives the (1,4) and leaves the (3,3).

Answer 2

[Walter Roberson]

If you have the image processing toolbox, you can use

bwboundary(I==8)

[Mohammad Golam Kibria]

your code gives me the following error:

??? Undefined function or method 'bwboundary' for input arguments of type 'logical'.

I don't know why? would you please explain me,what should I do?

Thanks

[Walter Roberson]

http://www.mathworks.com/help/toolbox/images/ref/bwboundaries.html

It appears you do not have the Image Processing Toolbox installed, or else you have a very old version. The Image Processing Toolbox is extra cost for all MATLAB editions except for the Student Edition.

[Sean de Wolski]

Typo Walter: bwboundaries(I==8)

Answer 3

[Ivan van der Kroon]

[rows,cols]=find(I==8)

rows =

     3
     2
     3
     4
     2
     1
     2

cols =

     2
     3
     3
     3
     4
     5
     5

Hope this helps.

[Mohammad Golam Kibria]

If my matrix becomes:

I =

1 1 1 1 8 1 2

1 1 8 8 8 2 1

1 8 8 8 8 1 1

1 1 8 8 8 1 1

2 2 2 1 1 1 1

2 2 2 1 1 1 1

In this case your code will not give proper output. It will give all location of 8

[Matt Fig]

What would be your expected result in this case?

[Walter Roberson]

Mohammad's expected result would exclude any of the 8's that are eight-connected only to other 8's: he wants only the 8's that are boundary pixels.

[Matt Fig]

I was wondering if Mohammad wanted 4 or 8 connected 8s excluded... I.e., in his example above, are there two 8s excluded or 1?

[Walter Roberson]

Hmmm, hard to say for sure, as his original example did not happen to include any 8's that were 4 connected to other 8's.

[Mohammad Golam Kibria]

I =

1 1 8 8 8 1 2

1 1 8 8 8 2 1

1 8 8 8 8 1 1

1 1 8 2 1 1 1

2 2 2 1 1 1 1

2 2 2 1 1 1 1

output needs the following:

(1,3) (1,5) (2,3) (2,5)(3,2)(3,3)(3,4)(3,5) (4,3)

perhaps this will help for better understanding my output

[Walter Roberson]

So to confirm, any "8" that has only 8's as neighbors is to be excluded, and all other 8's are to be included?

If the entire matrix was 2x2 and was

88

88

then everything should be excluded?

and for

881

888

881

then the entire left side is to be excluded?

Just include the 8's for which at least one of the neighbors is a non-8 ?

Answer 4

[Andrei Bobrov]

more variant

I1 = I == 8;
l1 = [ones(1,size(I1,2)); diff(I1)~=0];
l2 = [ones(size(I1,1),1)  diff(I1,1,2)~=0];
lud = [l1(2:end,:);false(1,size(I1,2))];
llr = [l2(:,2:end) false(size(I1,1),1)];
[j i] = find(((lud+l1+llr+l2)&I1)');
out = [i j];

more more variant

I1 = I==8;
I2=ones(size(I1)+2);
I2(2:end-1,2:end-1) = I1;
I1(conv2(I2,ones(3),'valid')==9)=0;
[i j] = find(I1);

without conv2

I1 = I==8;
I2=ones(size(I1)+2);
I2(2:end-1,2:end-1) = I1;
[i j] = find(I1);
ij1 = [i j];
ij2 = ij1 + 1;
IJ = arrayfun(@(x)bsxfun(@plus,ij2(:,x),[-1 0 1]),1:size(ij2,2),'un',0);
ij1(arrayfun(...
      @(x)sum(reshape(I2(IJ{1}(x,:),IJ{2}(x,:)),[],1)),1:size(ij2))==9,:) = [];

[Mohammad Golam Kibria]

I =

1 1 8 8 8 1 2

1 1 8 8 8 2 1

1 8 8 8 8 1 1

1 1 8 2 1 1 1

2 2 2 1 1 1 1

2 2 2 1 1 1 1

[i j] gives the following output

ans =

3 2

1 3

2 3

4 3

1 4

3 4

1 5

2 5

3 5

But I need the following output:

(1,3) (1,5) (2,3) (2,5)(3,2)(3,3)(3,4)(3,5) (4,3)

[Andrei Bobrov]

small correction in answer

[Andrei Bobrov]

more correction

Answer 5

[Teja Muppirala]

This will give all the 8's that are not entirely surrounded by other 8's. Assuming you have the Image Processing Toolbox.

[i,j] = find( (I==8) - imerode(I==8,ones(3)) )