# I have a basic level of image processing/matlab. I need to write an m file to complete skeletisation of a binary image of the word "skeleton" in white on a black background. It is to be done without using matlab functions for skeletising. thank you

10 views (last 30 days)
Ralph Dunne on 18 Apr 2013
Hi, I have a basic level of image processing/matlab. I need to write an m file to complete skeletisation of a binary image of the word "skeleton" in white on a black background. It is to be done without using matlab functions for skeletising. possibly one that will work on all binary images. Thanks
my code is as follows. I created it using an algorithm found online and adapted it to this:
clear all
I = im2double(I);
I = im2bw(I);
figure, imshow(I);
H = size(I, 1); %height of image
W = size(I, 2); %width of image
for i = 2:H-1
for j = 2:W-1
Neighbour = [I(i-1,j) I(i-1,j+1) I(i,j+1) I(i+1,j+1) I(i+1,j) I(i+1,j-1) I(i,j-1) I(i-1,j-1)];
Surrounds = [I(i-1,j) I(i-1,j+1) I(i,j+1) I(i+1,j+1) I(i+1,j) I(i+1,j-1) I(i,j-1) I(i-1,j-1) I(i-1,j)];
Transition = nnz(diff(Surrounds)==1);
Non_zero = sum(Neighbour(:)==1);
if Transition==1 && (2<=Non_zero<=6) && (I(i-1,j)*I(i,j+1)*I(i+1,j)==0) && (I(i,j+1)*I(i+1,j)*I(i,j-1)==0)
I(i,j)=0;
end
end
end
figure, imshow(I)

Ralph Dunne on 18 Apr 2013
which restriction? sorry I dont follow you.
Image Analyst on 18 Apr 2013
The restriction that you can't use the built-in MATLAB function that calculates the skeleton. I mean, why not use it?
Ralph Dunne on 18 Apr 2013
It is a project that I am completing and as part of the project no functions from matlab for the operation may be used. It must be built by creating my own m file which will complete the skeletising of the image. I am stuck from here I found this algorithm online at the link below under Parallel skeletonization algorithm 1 but I seem to have made a mistake as it does not fully work. It seems quite basic when explained in the article but I seem to have a problem with my coding if you have any ideas to help I would really appreciate it.

Thomas on 18 Apr 2013
The grassfire transform would be easy to implement: http://en.wikipedia.org/wiki/Grassfire_Transform
Be aware that this algorithm is not exceptionally fast. However, it can be parallelized.

#### 1 Comment

Ralph Dunne on 18 Apr 2013
thanks, I used the example algorithm found in the grassfire link you answered with. I am finding it hard to implement this algorithm. I am not a great coder. does the grassfire return a distance transfrom? the link below is the result I am trying to complete