2011/4/29 Tom Sharpless <[email protected]> > Whoever implements this should think about the following. > > Reducing image size properly requires low-pass filtering to prevent > aliasing. That is costly. But the image pyramid used for finding > interest points is already a low pass filter, and it will do about the > same work to bring the image to the same resolution. With the side > benefit that higher resolution 'scale images' will be available (in > case very sharp control points need to be located precisely) than if a > simple pre-reduction were done. > > So maybe the right way to get the speed benefit of 'reduction' is > simply to skip searching the higher resolution scale images for > interest points, at first, and to use them only later for 'polishing' > successful control point matches. Of course this assumes no memory > space problems :). > > Cheers, Tom > > > Hi Tom,
(Sorry for the late reply. I missed this one somehow). Thanks for sharing this with us. It's always good to have someone with enough background knowledge to put us on the right track. My interpretation from your answer is that you favor a parameter something like: "--skip_high_resolution_images" (or a better description) which will just do as you described and if the results are not good enough just rerun cpfind without the parameter. Or are you saying that cpfind should run first without the "high resolution images" and detect somehow (statistically?) that the results are not good enough and cpfind should rerun itself using the "high resolution images"? Harry -- You received this message because you are subscribed to the Google Groups "Hugin and other free panoramic software" group. A list of frequently asked questions is available at: http://wiki.panotools.org/Hugin_FAQ To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/hugin-ptx
