Extracting image coordinates from binary image

I'm new to matlab so apologies if this has been asked before.

I have a grayscsale image (of a tooth) that I need to extract the following data from;

Image centroid Max/Min x Max/Min y

I read the image into matlab and converted it to binary as below

im = 'C:\users\simon\desktop\w14.png' imread(im) bw = im2bw(imread(im),0.98)

inverted the data values so the tooth pixels have a value of 1 and the background is 0

Wrote the x and y values out by

[x y]=find([bw])

If i then take the mean of x and y, or use stats=regionprops(bw) to get the same and then plot the corresponding point back on the image it plots outside of the tooth.

Am i doing something wrong, or is this likely to be something wrong with the data itself?

Thanks!

Simon

Answer 1

[David Young]

Normally, one needs to write

[y, x] = find(bw);    % y before x (array ordering)

because the row indices are returned before the column indices. (Note that you don't need the extra [] round the argument.)

Then, provided there was only one non-zero region in bw, the means of x and y will plot at the centroid, using

plot(mean(x), mean(y), '*');    % x before y (image ordering)

However, you say you also have a problem with the results from regionprops. To go further, you may need to attach the image to your question, and show more of your code.

[Image Analyst]

Simon, you don't say "thanks" to David as your own "Answer" and then accept your own answer - David doesn't get any reputation points for that. So I've deleted your "Answer" moved it here since it's a reply to David rather than an "Answer" to your original question, and you now can accept his answer:

Thanks David - works perfectly - I hadn't realized that y was returned first (note to self - read manual first!)

P.S. if you want a tutorial on how to use regionprops() to find the centroid and weighted centroid of an image, see my Image Segmentation Tutorial in my File Exchange.

[David Young]

Thanks!

Answer 2

[Image Analyst]

Try this:

labeledImage = bwlabel(bw);
blobMeasurements = regionprops(labeledImage, 'Centroid');
% We can get the centroids of ALL the blobs into 2 arrays,
% one for the centroid x values and one for the centroid y values.
allBlobCentroids = [blobMeasurements.Centroid];
centroidsX = allBlobCentroids(1:2:end-1);
centroidsY = allBlobCentroids(2:2:end);
% Put the labels on the rgb labeled image also.
for k = 1 : numberOfBlobs           % Loop through all blobs.
      plot(centroidsX(k), centroidsY(k), 'b*', 'MarkerSize', 15);
      if k == 1
          hold on;
      end
      text(centroidsX(k) + 10, centroidsY(k),...
           num2str(k), 'FontSize', fontSize, 'FontWeight', 'Bold');
end

[SimonW]

Thanks ..

I have a bit of a follow up ...

Is there any way to do this on a 3D image stack to give the centroid as x,y,z? I guess I would need to loop through each image in turn?

I also need to get the Major and Minor Axes which regionprops gives for the stack in the x/y, x/z and y/z planes.

[Image Analyst]

Yep, you got it. To add measurements, list them in the arg list:

blobMeasurements = regionprops(labeledImage, 'Centroid',...
 'MajorAxisLength', 'MinorAxisLength');