Re: [hugin-ptx] Re: View this page "Affine SIFT algorith"
Anything that would help to spread control points farther beyond clumps in the center of the image for fisheye images... would greatly reduce stitching errors, I think. ei On Nov 7, 2010, at 2:58 AM, kfj wrote: On 31 Okt., 03:14, Tom Sharpless wrote: It is a fact that SIFT does not find as many matches as we would like in wide angle and especially fisheye images. But that problem will not be solved by better matching of affine transformations, as it is due to image deformations, characteristic of those lenses, that are not affine transformations. I've thought about the matter and come up with this train of thought: If you look at two corresponding small sections of, e.g., two fisheye images taken with different orientation, the transformation from one to the other is not, but resembles an affine transformation. The smaller the bits you compare, the closer the semblance. So looking at affine transformations helps, since a particular affine transformation locally approximates the actual nonaffine transformation. If the lens geometry is known, though, the choice of which affine transformation of which subsection of the image to feed into a feature generating algorithm could be narrowed down to that affine transformation that best locally approximates the non-affine transformation, thus gaining some efficiency. I admit that this is more of a gut feeling than something I could easily demonstrate mathematically, but maybe someone with a firmer mathematical background could give a comment? with regards Kay -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Re: View this page "Affine SIFT algorith"
ok im out for now, im going to check out some avidemux code...i think hugin can use some part of it and FFMPEG to read videos to convert them to diffrent projections and do video stitching...i opened a thread on this, havent seen any replies.. On Mon, Nov 1, 2010 at 11:43 AM, Bernd Hohmann wrote: > On 01.11.2010 16:49, john doe wrote: > > maybe its because the algorith isnt written for high detail pics, so it >> might work with 640*480 resolution images or 800*600... >> > > It works perfectly with high resolution images, its a matter of the number > of contrast changes (or so). I guess a fine checkerboard pattern will drive > it crazy :-) > > demo_ASIFT rescales to 800*600 internally. > > > Bernd > > -- > 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 hugin-ptx@googlegroups.com > To unsubscribe from this group, send email to > hugin-ptx+unsubscr...@googlegroups.com > For more options, visit this group at > http://groups.google.com/group/hugin-ptx > -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Re: View this page "Affine SIFT algorith"
On 01.11.2010 16:49, john doe wrote: maybe its because the algorith isnt written for high detail pics, so it might work with 640*480 resolution images or 800*600... It works perfectly with high resolution images, its a matter of the number of contrast changes (or so). I guess a fine checkerboard pattern will drive it crazy :-) demo_ASIFT rescales to 800*600 internally. Bernd -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Re: View this page "Affine SIFT algorith"
maybe its because the algorith isnt written for high detail pics, so it might work with 640*480 resolution images or 800*600... On Mon, Nov 1, 2010 at 10:48 AM, Bernd Hohmann wrote: > On 01.11.2010 14:32, Tom Sharpless wrote: > > Tom, > > > As far as I can see, ASIFT is doing its job very well even with fisheye >>> images and without calibration data. >>> >>> That is surprising to me. As I understand the description of ASIFT >> (which is not very well) it assumes that the source images are in >> rectilinear projection. >> > > I don't think that the projection of the original images does matter > because the algorithm bend and distort them in all directions. > > > However, the results should speak for themselves. >> > > Indeed. > > I created a testcase: 8 fisheye shots, all of them "shots from the hip". > > All the cp finders included in hugin went wrong, ASIFT created enouth good > points so that all images were almost in the right position. > > On images with high detail (for example my shots of an ancient lime rock > mine), the algorithm gets painful slow. About 90 seconds for generating the > keypoints and about 20-25 minutes for matching. > > Bernd > > > -- > 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 hugin-ptx@googlegroups.com > To unsubscribe from this group, send email to > hugin-ptx+unsubscr...@googlegroups.com > For more options, visit this group at > http://groups.google.com/group/hugin-ptx > -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Re: View this page "Affine SIFT algorith"
On 01.11.2010 14:32, Tom Sharpless wrote: Tom, As far as I can see, ASIFT is doing its job very well even with fisheye images and without calibration data. That is surprising to me. As I understand the description of ASIFT (which is not very well) it assumes that the source images are in rectilinear projection. I don't think that the projection of the original images does matter because the algorithm bend and distort them in all directions. However, the results should speak for themselves. Indeed. I created a testcase: 8 fisheye shots, all of them "shots from the hip". All the cp finders included in hugin went wrong, ASIFT created enouth good points so that all images were almost in the right position. On images with high detail (for example my shots of an ancient lime rock mine), the algorithm gets painful slow. About 90 seconds for generating the keypoints and about 20-25 minutes for matching. Bernd -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Re: View this page "Affine SIFT algorith"
so, it would be an interesting thing to create a template.pto before, and include it in the ASIFT source code, and see what happens?? On Mon, Nov 1, 2010 at 9:02 AM, Tom Sharpless wrote: > Hi Bernd > > On Oct 31, 7:02 am, Bernd Hohmann wrote: > > On 31.10.2010 03:14, Tom Sharpless wrote: > > > > Tom, > > > > > It is a fact that SIFT does not find as many matches as we would like > > > in wide angle and especially fisheye images. But that problem will > > > not be solved by better matching of affine transformations, as it is > > > due to image deformations, characteristic of those lenses, that are > > > not affine transformations. > > > > As far as I can see, ASIFT is doing its job very well even with fisheye > > images and without calibration data. > > > That is surprising to me. As I understand the description of ASIFT > (which is not very well) it assumes that the source images are in > rectilinear projection. I do agree with Morel and Yu's idea that > simulating rotations of the direction of view should improve local > similarity measured at true match points. However the results of that > rotation do depend on the lens projection function, so I would not > expect such a good result when the wrong lens function is used. > However, the results should speak for themselves. > > In panoramic image sets, the images are in fact related by pure > rotations of the view direction (leaving aside parallax errors). > Therefore, if lens distortion is corrected, the Morel/Yu procedure > should generate locally almost identical images, when the chosen > rotations happen to equal the true ones. By using the approximate > alignment angles from a "shooting template", it should be possible to > limit the range of angles that needs to be searched; and of course > also to limit the image areas that need to be searched. This could > lead to a rather fast alignment algorithm. > > Regards, Tom > > > Bernd > > > > -- > > Bernd Hohmann > > Dipl. Organisationsprogrammierer > > DV Sachverständiger & Gutachter > > Höhenstrasse 2 * 61130 Nidderau > > Telefon: 06187/900495 * Telefax: 06187/900493 > > Blog:http://blog.harddiskcafe.de > > -- > 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 hugin-ptx@googlegroups.com > To unsubscribe from this group, send email to > hugin-ptx+unsubscr...@googlegroups.com > For more options, visit this group at > http://groups.google.com/group/hugin-ptx > -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Re: View this page "Affine SIFT algorith"
On 31.10.2010 19:25, john doe wrote: yes your wrapper, do you have a program to show the code working?? I beg your pardon, but my english is not good enough to understand you. The wrapper is a simple bash script and what it does is documented in the sourcecode (and it does not very much except for converting files from one format to another format). btw: I finished 1.01 which condenses the controlpoints more (or less). http://201.ag/hugin/asiftw MD5: 65a0bcc63d5e82f1548a668bd33abf99 Bernd -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Re: View this page "Affine SIFT algorith"
yes your wrapper, do you have a program to show the code working??if so can you upload some pics of it?thanks.. On Sun, Oct 31, 2010 at 1:30 PM, Bernd Hohmann wrote: > On 31.10.2010 18:51, john doe wrote: > > hey bernd hows everything going with the coding??can you supply images >> as how the code works??including aligning images with ASIFT if possible... >> > > Which coding? My wrapper? Images for what? > > Bernd > > > -- > 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 hugin-ptx@googlegroups.com > To unsubscribe from this group, send email to > hugin-ptx+unsubscr...@googlegroups.com > For more options, visit this group at > http://groups.google.com/group/hugin-ptx > -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Re: View this page "Affine SIFT algorith"
On 31.10.2010 18:51, john doe wrote: hey bernd hows everything going with the coding??can you supply images as how the code works??including aligning images with ASIFT if possible... Which coding? My wrapper? Images for what? Bernd -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Re: View this page "Affine SIFT algorith"
hey bernd hows everything going with the coding??can you supply images as how the code works??including aligning images with ASIFT if possible... On Sun, Oct 31, 2010 at 6:32 AM, Bernd Hohmann wrote: > On 31.10.2010 03:14, Tom Sharpless wrote: > > Tom, > > > It is a fact that SIFT does not find as many matches as we would like >> in wide angle and especially fisheye images. But that problem will >> not be solved by better matching of affine transformations, as it is >> due to image deformations, characteristic of those lenses, that are >> not affine transformations. >> > > As far as I can see, ASIFT is doing its job very well even with fisheye > images and without calibration data. > > Bernd > > -- > Bernd Hohmann > Dipl. Organisationsprogrammierer > DV Sachverständiger & Gutachter > Höhenstrasse 2 * 61130 Nidderau > Telefon: 06187/900495 * Telefax: 06187/900493 > Blog: http://blog.harddiskcafe.de > > > -- > 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 hugin-ptx@googlegroups.com > To unsubscribe from this group, send email to > hugin-ptx+unsubscr...@googlegroups.com > For more options, visit this group at > http://groups.google.com/group/hugin-ptx > -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx
Re: [hugin-ptx] Re: View this page "Affine SIFT algorith"
On 31.10.2010 03:14, Tom Sharpless wrote: Tom, It is a fact that SIFT does not find as many matches as we would like in wide angle and especially fisheye images. But that problem will not be solved by better matching of affine transformations, as it is due to image deformations, characteristic of those lenses, that are not affine transformations. As far as I can see, ASIFT is doing its job very well even with fisheye images and without calibration data. Bernd -- Bernd Hohmann Dipl. Organisationsprogrammierer DV Sachverständiger & Gutachter Höhenstrasse 2 * 61130 Nidderau Telefon: 06187/900495 * Telefax: 06187/900493 Blog: http://blog.harddiskcafe.de -- 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 hugin-ptx@googlegroups.com To unsubscribe from this group, send email to hugin-ptx+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/hugin-ptx