Estimating spatial frequency

[SansGrip]

I've been looking for a simple explanation of spatial frequency and how to estimate it efficiently. I asked Dr Google but he must've been having a bad day because all he found was tons and tons of very complicated math ;).

Basically what I want to do is split an image up into blocks and estimate (or calculate precisely, if it can be done fast enough) the spatial frequency of each one. I realize this is sort of what MPEG encoders do but I thought I'd ask here before jumping into the XviD source...

I'm currently wading through the FFT section of Numerical Recipes in C, but it's giving me headaches. Can anyone advise me, other than to stock up on Aspirin? :D

[MfA]

It depends on the context, but because you ask here I assume it is in relation to the human visual system. Look up the cortex transform on google ... that should give you something usefull.

If you are interested in spatial frequencies for using spatial frequency sensitivity in coding be warned ... intra-channel masking is only moderately usefull, and inter-channel masking in natural images is too complex for a model to be determined from simple grating experiments (there are no good models for it in fact AFAIK, one might even question wether overall spatial channels are the best approach for these kind of applications).

[SansGrip]

If you are interested in spatial frequencies for using spatial frequency sensitivity in coding be warned ... intra-channel masking is only moderately usefull, and inter-channel masking in natural images is too complex for a model to be determined from simple grating experiments

I have absolutely no idea what this means, but thanks for trying :).

In the end I found a method of estimating it that seems to work pretty well for my purposes.