DocumentCode :
2400632
Title :
Approximate earth mover’s distance in linear time
Author :
Shirdhonkar, Sameer ; Jacobs, David W.
Author_Institution :
Center for Autom. Res., Univ. of Maryland, College Park, MD
fYear :
2008
fDate :
23-28 June 2008
Firstpage :
1
Lastpage :
8
Abstract :
The earth moverpsilas distance (EMD) is an important perceptually meaningful metric for comparing histograms, but it suffers from high (O(N3 logN)) computational complexity. We present a novel linear time algorithm for approximating the EMD for low dimensional histograms using the sum of absolute values of the weighted wavelet coefficients of the difference histogram. EMD computation is a special case of the Kantorovich-Rubinstein transshipment problem, and we exploit the Holder continuity constraint in its dual form to convert it into a simple optimization problem with an explicit solution in the wavelet domain. We prove that the resulting wavelet EMD metric is equivalent to EMD, i.e. the ratio of the two is bounded. We also provide estimates for the bounds. The weighted wavelet transform can be computed in time linear in the number of histogram bins, while the comparison is about as fast as for normal Euclidean distance or chi2 statistic. We experimentally show that wavelet EMD is a good approximation to EMD, has similar performance, but requires much less computation.
Keywords :
computational complexity; image matching; wavelet transforms; Holder continuity constraint; Kantorovich-Rubinstein transshipment problem; computational complexity; earth movers distance; histograms; linear time algorithm; normal Euclidean distance; weighted wavelet transform; Computational complexity; Earth; Extraterrestrial measurements; Histograms; Image retrieval; Jacobian matrices; Pattern matching; Statistics; Wavelet coefficients; Wavelet domain;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Computer Vision and Pattern Recognition, 2008. CVPR 2008. IEEE Conference on
Conference_Location :
Anchorage, AK
ISSN :
1063-6919
Print_ISBN :
978-1-4244-2242-5
Electronic_ISBN :
1063-6919
Type :
conf
DOI :
10.1109/CVPR.2008.4587662
Filename :
4587662
Link To Document :
بازگشت