powerspectrum.m

Get the normalized power spectrum of an image.

The power spectrum measures how much information is contained at each spatial frequency.

Construction of the power spectrum takes three steps:

1) Get the 2D (fftshifted) power spectrum
2) Sample lines through sf=0 and rotated by 1degree intervals
3) Average sampled lines to get result

This function works fast compared with other approaches I've tried, the most obvious (and worst) of which is to rotate the entire 2D spectrum.

Finally, this function can accurately sample images with nonsquare pixels (such as in some medical imaging applications).