next up previous contents index
Next: ll_boundaries2 Up: Reference Previous: canny   Contents   Index

harris

$ \bigcirc$Name


harris Harris corner detector




$ \bigcirc$Command Synopsis


harris [-k k] [-g g] [-s s] [-t t] in out



-k k : weight of trace, default 0.04 (suggested by Harris)

-g g : standard deviation of gaussian filtering, default 1.0

-s s : size of neighborhood for maxima research

-t t : log10(threshold) for cornerness, default 4.0

in : Input image

out : Corner list (Flist)




$ \bigcirc$Function Summary


Flist harris (in , k , g , size , t )

Fimage in ;

float *k , *g ;

double *t ;

int *size ;




$ \bigcirc$Description


This module is an implementation of the classical corner detector proposed by Harris and Stephen [HS88]. It computes a cornerness function defined by

C = det M - k(tr M)2,

where M is the 2×2 matrix g$\scriptstyle \sigma$*(Du . Dut) computed from the image gradient Du and a smoothing with a gaussian kernel with standard deviation $ \sigma$. The value k = 0.04 has been proposed by the authors of [HS88]. Corners are local maxima of this cornerness function. The size of the neighborhood for maxima computation and a threshold for their value have to be fixed. The output is a Flist structure containing the coordinates of the detected corners.




$ \bigcirc$See Also


fderiv, fsepconvol.




$ \bigcirc$Version 1.0


Last Modification date : Fri Apr 11 09:10:57 2003


$ \bigcirc$Author


Frederic Cao






next up previous contents index
Next: ll_boundaries2 Up: Reference Previous: canny   Contents   Index
mw 2004-05-05