Brian Moore

@Jean Bilheux: The algorithm is pretty simple: move a small arc length, say ds, along the border of each shape and then rotate/translate the inner shape as necessary to align the tangent vectors of the inner/outer shapes at their current border positions

I.e., all one needs is an expression for the derivative of each shape's border, from which one can compute arc length (via numerical integration, if necessary) and tangent vectors

Sure, try running it like this:

imageStr = 'lena_color.png'; % image name
iter = 2; % number of wavelet iterations
cf = 0.94; % compress 94% of the wavelet coefficients
type = 'D6'; % use Daubechies D6 wavelets
PStageWavelets(imageStr,iter,cf,type);

This will compute the Daubechies 'D6' wavelet coefficients (iterated twice) of the color image 'lena_color.png', zero out the smallest 94% of the wavelet coefficients in absolute value, and reconstruct the original image from the compressed coefficients.

Hope this helps!

This is great! I had totally forgotten about spirograph, had one when I was a kid. Nice implementation in MATLAB and fun to play around with :)

@Jean Bilheux: The algorithm is pretty simple: move a small arc length, say ds, along the border of each shape and then rotate/translate the inner shape as necessary to align the tangent vectors of the inner/outer shapes at their current border positions

I.e., all one needs is an expression for the derivative of each shape's border, from which one can compute arc length (via numerical integration, if necessary) and tangent vectors

Cool!