Clarification of idx in Planar Object Measuring tutorial

Under the "Detect Coins" paragraph in the "Measuring Planar Objects using a calibrated camera" example, the lines:

[areas, boxes] = step(blobAnalysis, imCoin);

% Sort connected components in descending order by area
[~, idx] = sort(areas, 'Descend');

% Get the two largest components.
boxes = double(boxes(idx(1:2), :));

% Adjust for coordinate system shift caused by undistortImage
boxes(:, 1:2) = bsxfun(@plus, boxes(:, 1:2), newOrigin);

I am uncertain of what the brackets, after the second occurrence of idx, are trying to define. I am trying to apply this to my own image and it gives me an error which states: Index exceeds matrix dimensions. Could someone please clarify this for me?

[Guillaume]

Could you provide a link to the tutorial you mention? You can right click on the page in the help browser and select Get Page Address.

[Kasia Hearn]

http://www.mathworks.com/help/vision/examples/measuring-planar-objects-with-a-calibrated-camera.html

Answer 1

[Guillaume]

After the sort idx contains the indices of the areas vector sorted in descending order.

The following, idx(1:2), which could also have been written as idx([1 2]), just take the first two indices. This is used to get the two rows of boxes that correspond to the two greatest area. The obvious reason why you'd get an index exceeds matrix dimension error would be because there are less than two areas because only one object (or none) was detected.

The next line just add newOrigin to columns 1 and 2 of the two selected boxes.

[Kasia Hearn]

Thank you for your answer. I am trying to detect the coordinates of the sun in an image I've taken. I just want to know if you think its possible by following this tutorial?

Thank you.