finding distance between centroids

Hi

I have a group of blobs of a binary image for which I have calculated the centroids.

I would like to find the distance between each of these centroids and then extract the two centroids which lie in a 40 to 45 range. I've seperated the X and Y cdts into 2 matrices like so:

>> X=centroids(:,1); Y=centroids(:,2); >> X

X =

121.3073
142.9425
168.1000
225.5856
522.9487

>> Y

Y =

207.2626
149.0276
114.7667
297.5586
299.9551

and then I tried computing the euclidean distance using pdist2 but I dont understand what it gives me and I dont see any 40s over here

D = pdist2(X,Y,'euclidean'); >> D

D =

   85.9553   27.7203    6.5406  176.2513  178.6479
   64.3200    6.0851   28.1759  154.6160  157.0126
   39.1626   19.0724   53.3333  129.4586  131.8551
   18.3230   76.5580  110.8189   71.9730   74.3695
  315.6861  373.9211  408.1821  225.3902  222.9936

Any advice pls?

Answer 1

[Andrei Bobrov]

try

D = dist([X,Y]')

[mayfair]

Thanks andrei I just tried it and got this:

D = dist([X,Y]')

D =

0 62.1241 103.6583 137.9396 412.1987

62.1241 0 42.5054 169.9745 408.8812

103.6583 42.5054 0 191.6180 400.2654

137.9396 169.9745 191.6180 0 297.3728

412.1987 408.8812 400.2654 297.3728 0

I realised that dist and pdist2 actually computes distance between each X and Y which is not exactly what I want...

I'd like to be able to find the distance between each set of centroid coordinates and ideas on that pls?

Thanks!

[Ashish Uthama]

mayfair, try hand computing your expected result for a couple and compare with the above.

Example: Distance betwee first two points:

>> ( (X(1)-X(2))^2 + (Y(1)-Y(2))^2)^.5

ans =

62.1240

(Isnt that what Andrei's answer gives?)

[mayfair]

That's truue Ashish I managed to work it out I found that if you subtract the x and y matrices from its transposed matrix then use the hypot function it works like a charm :)

[Rujal]

hi mayfair..

My question is same as yours please give me a matlab code to find out distance between set of centroids. Thanks in advance.

[Image Analyst]

If you have the stats toolbox, you can use pdist2(). If not, just a simple for loop with vectorized distance calculation between that point and all the other points will do it.

[Guillaume]

Even without the stats toolbox, you don't need a loop:

euclideandistance = hypot(bsxfun(@minus, X, X'), bsxfun(@minus, Y, Y'));

is all that's needed (assuming X and Y are vectors of the same size)

[emami.m]

Good job Guillaume,

It was a smart and exact solution.