How to find the nearest object to a reference point in binary image?

Dear all,

I have a binary image that contains two objects(Object_A, Object_B) very close to each other as shown below. I want to find which object is closer to a reference point in order to be selected (Object_B in this example).

I was thinking to find the distance of the centroid of each object to the reference point (in this example the centroid of the reference point X = 360, Y = 330), but this will not work because in few cases the centroid of the far object could be closer to the reference point as I tried to explain in the figure below.

Is there any way to find the closest object to a reference point?

Any help and suggestion will be appreciated.

Meshoo

Answer 1

[Image Analyst]

It's trivial. Just use bwboundaries to get the list of (x,y) coordinates of the boundary and then put in a for loop, calculating the distance using the Pythagorean Theorem to get the boundary that is the closest. In the for loop, you can vectorize it and use min().

boundaries = bwboundaries(binaryImage);
overallMinDistance = inf;
for b = 1 : length(boundaries)
   thisBoundaryX = boundaries{b}(:, 1);
   thisBoundaryY = boundaries{b}(:, 2);
   distances = sqrt((thisBoundaryX - refx).^2 + (thisBoundaryY - refy).^2);
   [minDistance, indexOfMin] = min(distances);
   if minDistance < overallMinDistance
      % This boundary is the closer one.
.............you finish the rest - it's easy.
end

[Meshooo]

Thank you very much Image Analyst. I think it is similar to one old post that you answered some time ago. I tried but didn't work and I wrote a comment in your answer:

http://www.mathworks.com/matlabcentral/answers/152281-calculate-the-distance-between-two-objects-automatically-using-matlab

Is there any easier way to do that?

Thank you.

[Image Analyst]

No - this is as easy as it gets. Let me see your code. It should be really trivial so you probably just made a simple mistake.