Calculate percentage of similarities in two polygons

8 views (last 30 days)
Hello friends.
Is there any framework in Matlab that can calculate the percentage of similarities between two polygons?
If not, how can i implement one?
Googling i found out a formula to calculate the similarities of two polygons R and P2 using this formula:
But i don't know how to calculate the union and intersection polygons.
Any help please? :)
Thanks in advance!

Accepted Answer

Kelly Kearney
Kelly Kearney on 10 Nov 2015
If you have the Mapping Toolbox and want a more exact answer:
% Some example polygons
theta = linspace(0, 2*pi, 100);
x1 = cos(theta) - 0.5;
y1 = -sin(theta);
x2 = x1 + 1;
y2 = y1;
% Calculate
[xint, yint] = polybool('&', x1, y1, x2, y2);
[xun, yun] = polybool('|', x1, y1, x2, y2);
a1 = polyarea(xun, yun);
a2 = polyarea(xint, yint);
sim = 1 - (a1- a2)./a1;
  1 Comment
Amaral
Amaral on 10 Nov 2015
Edited: Amaral on 10 Nov 2015
It worked very well. It gives accurate results which is exactly what i'm looking for. Thanks :)

Sign in to comment.

More Answers (1)

Image Analyst
Image Analyst on 9 Nov 2015
If that definition works for you, then that's fine. One could think of others, but whatever works for what you need it to, is fine. I think the easiest way to solve the formula is to convert the polygons to images with poly2mask().
rows = 1920; % Whatever is needed to contain your points.
columns = 1600;
image1 = poly2mask(x1,y1, rows, columns);
image2 = poly2mask(x2,y2, rows, columns);
unionImage = image1 & image2;
intersectionImage = xor(image1, image2)
similarity - 1 - (sum(unionImage(:) - intersectionImage(:)) / sum(unionImage(:));
If your x and y are too small, then multiply them by some scaling factor beforehand, like 100 or 1000 or whatever so that your image is big and not too pixelated.
  2 Comments
Amaral
Amaral on 9 Nov 2015
Thanks for the answer. :) I'll try it out :)
Amaral
Amaral on 9 Nov 2015
Edited: Amaral on 9 Nov 2015
Your approach worked. But had to do some adjustments because i think the way you calculated the union and intersection is wrong. This way it gives the results i want, at least i think so according to the results in the tests :P
unionImage = image1 | image2;
intersectionImage = image1 & image2;
similarity = (1 - (sum(unionImage(:) - intersectionImage(:)) / sum(unionImage(:)))) * 100

Sign in to comment.

Categories

Find more on Images in Help Center and File Exchange

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!