[ptx] Enblend Multiresolution Spline Blender

Andrew C Mihal mihal at eecs.berkeley.edu
Mon Mar 15 07:59:29 GMT 2004


On Fri, 12 Mar 2004, Pablo d'Angelo wrote:

> initalize pano with first image + mask + pyramids.
> 
> for images
>   // create masks for blending
>   createMask(panoMask,image);
>   // build laplacian pyramid for current image
>   buildPyramids(image, imageLP);
>   blend(panoLP, imageLP);
>   // update panoMask, as well
> 
> collapse(panoPyr)
> 
> I hope it is possible to move the collapse out of the loop.

As to this last point, unfortunately I don't think that it is 
mathematically sound. The layers of the laplacian pyramid do not form an 
orthogonal basis set for the image. Say that after a blending step you 
introduce a new smoothness into the image on a particular layer. That 
smoothness may acutally belong on a lower layer. If you collapse and then 
reexpand the pyramid, then future blends will use the fullest transition 
region possible against that feature. Otherwise they will use a suboptimal 
transition region. In practice I don't know if you'll ever notice a 
difference, but at least mathematically you should collapse and expand 
after each blend step.

I've replaced the erosive thinning algorithm for mask creation - next I'm 
going to try to calculate a tighter ROI. That and a couple other changes 
to save memory and it'll be version 1.0. New features will come after 
that.

Andrew

---------------------------------
Andrew Mihal
www-cad.eecs.berkeley.edu/~mihal
mihal at eecs.berkeley.edu



More information about the ptX mailing list