om,

I've been playing around with an immage pre-processor that is intended
to mimic the center-surround pre-processing of the human eye...

My first attempt worked on a 3x3 matrix adding the brightness of the
pixels multiplied by the factors as follows:

1/12 1/6 1/12
1/6 -1 1/6
1/12 1/6 1/12

This algorithm had the following problems:

1. it's range was too small, features in the source immage that change
over several pixels were not detected or detected poorly.

2. multiplying the result or making the result proportional to total
brightess made the algorithm too sensitive to subtle features but not
sensitive to larger ones....

Changing the grid to:

0 0 1/6 0 0
0 1/12 0 1/12 0
1/6 0 -1 0 1/6
0 1/12 0 1/12 0
0 0 1/6 0 0

improved matters but tends to blur the immage so that smaller details
are lost and complex edges are blurred...

Has anyone here worked on this kind of problem before? do you have any
suggestions?

--
President Bush's *head* is _F_L_A_T_.

http://users.rcn.com/alangrimes/