I am looking for a common distance method to compare two distribution (ex: histogram of image). Please suggest to me some common method to do it. I found some method ex: Bhattacharyya distance , K-L distance. Do you have other?
3
-
$\begingroup$ mathoverflow.net/questions/103115/… $\endgroup$– UwFCommented Sep 13, 2014 at 7:52
-
$\begingroup$ try earthmover distance:en.wikipedia.org/wiki/Earth_mover's_distance $\endgroup$– Aryeh KontorovichCommented Sep 14, 2014 at 4:30
-
$\begingroup$ @AryehKontorovich: It is very good answer. Thank you so much. I have read it. But it already published in some paper. So i need find other solution. Do you know other solution for this problem. Some papers that I found as ieeexplore.ieee.org/stamp/stamp.jsp?arnumber=6116341 and ieeexplore.ieee.org/xpl/… $\endgroup$– JohnCommented Sep 14, 2014 at 15:33
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
NID,normalized information distance which is motivated by Kolmogorov complexity ,please see papers of Ming Li's Homepage http://homepages.cwi.nl/~paulv/learning.html or Paul Vitaniy's homepage http://homepages.cwi.nl/~paulv/learning.html.In fact,the articles are the same.
Somehow,it is slow when you run the program of it ,and it is just a approximate algorithm since Kolmogorov complexit is non computable.Actually,it is a univesal distance,which is the best among all computable distance.
answered Sep 13, 2014 at 7:02