MATLAB and Simulink resources for Arduino, LEGO, and Raspberry Pi

# How to convert binary image to graph?!

Asked by Mariam Sheha on 1 Jul 2013

Hi every body...

Hi everybody,

Can you please help me to convert binary image results from Laplacian of Gaussian (LOG) filter. to graph?!

as that attached figure?! http://imageshack.us/photo/my-images/41/sww.png/

Thanks A lot.....

http://imageshack.us/photo/my-images/41/sww.png

## Products

Answer by Image Analyst on 1 Jul 2013

Neither do I. You'd have to supply more information. Why do you want to do that anyway?

Mariam Sheha on 1 Jul 2013

let's assume that we got the connecting component of the binary image through (cc = bwconncomp(bw, 4);) then i wana to graph those connected points using that function ([S, C] = graphconncomp(S);)... how can i do that?!...

About what i want to do, Look i am trying to detect some cyclic structure and meshes present in am image through different structure operations , simple representation for work structure represented at that image (stopped at : the convertion from binary to graph step): http://imageshack.us/content_round.php?page=done&l=img208/1244/tdh1.png

Image Analyst on 1 Jul 2013

It's not clear from the figure. You'd have to read the paper more carefully.

Answer by Youssef KHMOU on 1 Jul 2013

hi,

You can convert the image to graph by simply plot the result as vector or matrix :

``` I=imread('circuit.tif');
figure, plot(I);
figure, plot(I(:)), title('  Mapping operator');
BB=im2bw(I);
figure, plot(BB);
F=sparse(BB);
figure, spy(F);```

Many other possibilities do exist, try those above and see .

Image Analyst on 2 Jul 2013

The graph seems like it used random locations from only the boundary. I have no idea why -- but she may need to call bwboundaries() or bwperim() to get the boundary.

Mariam Sheha on 3 Jul 2013

Thanks A lot for answering me...

I think that "Youssef Khmou" could be secound step after using function recommended by "Image Analyst" to get the boundary of each white region...

So , for the conversion step from binary to graph i suppose to treat each region as anode and connect it to the other near node, that's why i think "Image Analyst" answer is the accepted were (i may get the boundary of each region and it's centroid then connect between them consider the centriod being a node) that what i moving through now...

Thanks Image Analyst & Youssef Khmou, and please if you have any idea about what i am saying replay me..

Kind Regards;

Image Analyst on 3 Jul 2013

It looks like the log filter is just one big region. Not sure why there are multiple regions. Then they somehow get random (or so it appears) points from the perimeter and then the next step is somehow they have multiple small regions outlined in green. I have no idea how they went from the log image to the graph to the cyclic subgraphs (multi-region) images. Maybe you do. Why are there multiple regions?