Data for EMD papers

These are the datasets used in our EBDT 2006 and ICDE 2006 papers on the earth mover's distance (EMD). See my list of publications for citations.

The first line of each histogram file contains three numbers: The height (8) and width (12) of the image (measured by the number of tiles), and the number of dimensions of the feature (12 for the color layout descriptor). The remaining lines are the color layout features. There are 8 x 12 = 96 of these lines, one for each tile in row-major order (English reading order). Each line has 12 numbers, one for each dimension of the feature.

Last updated: 2007-06-09