Matrix formula of cornerEigenValsAndVecs different from the one harrisCorner proposed

asked 2013-09-09 14:46:52 -0600

stereomatching gravatar image

updated 2013-09-10 01:01:34 -0600

Moster gravatar image

According to this page cornerEigenValsAndVecs, the M of cornerEigenValsAndVecs are different than the harrisCorner proposed.

(1)    (dI / dx) * (dI / dx) == A
(2)    (dI / dx) * (dI / dy) == B
(2)    (B * B) == C
(3)    (dI / dy) * (dI / dy) == D

The M of cornerEigenValsAndVecs on the link is

A C
C D

The M of harrisCorner proposed is

A B
B D

It is a mistake of the document?I get it wrong?Or there are a reason to calculate the M like the document say?

edit retag flag offensive close merge delete