Revision history [back]

it's the chi-sqr distance not the distribution here, which matters, and you use it like:

double dist = compareHist(histo_A, histo_B, HISTCMP_CHISQR);

again, have a look at the opencv code