Anyway to count poker chip stacks?

Hi, does anyone have an idea on how to count, how many stacks there are in this image, or even better, count how many chips. I've been searching all around the forums here, and still didnt get a way to do this.

Thanks.

Answer 1

[Image Analyst]

Have columns where each stack may start and stop, like 10 of them for 5 possible stacks. Then extract each column and take the min horizontally. Then call find() to find out where the top and bottom of each stack is. Measure the height divide by the known height of a single chip. Then you need to take special care for when there are no chips at all - topTow will be empty in that case. Here is some code to get you started. Finish it, and/or come back with modified code if you still need help.

rgbImage = imread(filename);
grayImage = min(rgbImage, [], 3);
% Define columns(5, 2) for starting and stopping columns for each of the 5 possible places.
for k = 1 : 5
    thisStack = grayImage(:, columns(k,1):columns(k,2));
    verticalProfile = min(thisStack, [], 2);
    topRow = find(verticalProfile < someValue, 1, 'first');
    bottomRow = find(verticalProfile < someValue, 1, 'last');
    numberOfChips(k) = (bottomRow-topRow) / rowsForOneChip;
end

[Rafael Peres]

Hi, when you say define columns, you're saying define the columns "manually" on top of the stacks, and then divide the size by the size of the chip? If that is what i understand, is a nice approach for getting this done, but i would like to identify the stacks automatically, that is what i wasn't able to do so far. Sorry if i understand you wrong. And if i can identify the stacks automatically your column approach will be of great help.

Thank you!

[Image Analyst]

If you will have images that you have no control over with stacks all over the place, obscuring other stacks, with different backgrounds, etc., like this:

then you're in for a ton of work. I was assuming you would have your camera setup in a known position with a known field of view, in which case you can definitely say in advance which columns might contain a stack. In essence, you're defining a template. If you have images like the above, then good luck to you.

[Rafael Peres]

The idea is to have images like the one in the first quote, the stacks aligned in a front position taken picture, with no obscuring stacks in front of the others. I asked if i'll have to place the columns myself, or if there is a way to identify the form of stacks, like we do with circles, rectangles and other forms, if i can get this done with your column approach, it would be perfect. I'm sorry if misunderstood your first answer.

Thanks again.

[Image Analyst]

Rafael, it was quicker just to do it for you. See attached demo.

[Rafael Peres]

Oh, thank you very much, sorry again, image processing isn't my area, i'm just starting to learn something with the tutorials given by matlab.

[Image Analyst]

You're welcome. If an Answer solves your question, you can click "Accept this answer" to give the answerer "credit" for the help.