hessian matrix

hi all:

i need to calculate the characteristcs points of two images so i can
thereafter register them.

i've been told to calculate for each point the hessian matrix and
then get their eigenvalues. if the minor eigenvalue is greater than a
threshold then the point is a characteristic point....

my problem.... don't know how to calculate an hessian..

i've wrote some code to make an approximation to the derivate
definition.

f'(x) = lim (f(x+h) - f(x))/h
h->0

for instance, my function's input is a vector. so to calculate the
derivate of that vector i circshift by h = 1 that vector and then
subtract the original on. as h = 1 i don't need to make any
division.... after i only have length(vector)-1 derivate values....

i suspect that this is not the best way to calculate a discrete but
for now it was the only thing i could remember.

with the derivate function calculate i would do the same to get the
second order derivate.... and so on until i got the hessian

if any one has any ideas..... please tell me...

thanks to all

mut
.