Hyperstack registration
Locked
 
|
Hello all,
I have a hyperstack (2-channels, 130 z-slices, 256x256 pixels). I have two issues with the data that I need to address. 1) There was drift during the experiment and there are small translations between most z-slices 2) There was one larger jump during the experiment. If I only operate on one channel, StackReg does a good job of removing the slow drift, but doesn't handle the larger jump (~30-50 pixels). I would also like the second channel to be shifted the same as the first channel, but not separately registered. So my two questions are: 1) Is there a good way to manually shift all the slices after the jump so StackReg can do its job? 2) Is there a way to then do the exact same shifts on a parallel stack (or another channel in a hyperstack)? Thanks, --David |
Re: Hyperstack registration
|
|
Hi David,
did you try the "Plugins > Registration > Descriptor-based series registration (3d+t)" which is included in Fiji, it might be able to do the job automatically. It can handle arbitrary jumps and usually aligns long time series reliably. It supports 3d-translation, 3d-rigid and 3d-affine (in 2d also 2d-similarity and 2d-perspective). It uses local features in the image stack to achieve that. I did not have time to write the documentation yet, it uses a similar algorithm as the SPIM registration but uses the image content itself instead of beads: http://fiji.sc/wiki/index.php/SPIM_Registration Nice greetings, Stephan On Feb 24, 2012, at 15:06 , David Hovis wrote: > Hello all, > > I have a hyperstack (2-channels, 130 z-slices, 256x256 pixels). I have two issues with the data that I need to address. > > 1) There was drift during the experiment and there are small translations between most z-slices > 2) There was one larger jump during the experiment. > > > If I only operate on one channel, StackReg does a good job of removing the slow drift, but doesn't handle the larger jump (~30-50 pixels). > > I would also like the second channel to be shifted the same as the first channel, but not separately registered. > > So my two questions are: > 1) Is there a good way to manually shift all the slices after the jump so StackReg can do its job? > 2) Is there a way to then do the exact same shifts on a parallel stack (or another channel in a hyperstack)? > > Thanks, > > --David |
|
|
In reply to this post by David Hovis
You could try the Image Stabilizer plugin. It will log translations and
apply the same ones again when requested: http://www.cs.cmu.edu/~kangli/code/Image_Stabilizer.html On Fri, Feb 24, 2012 at 10:06 AM, David Hovis <[hidden email]> wrote: > Hello all, > > I have a hyperstack (2-channels, 130 z-slices, 256x256 pixels). I have > two issues with the data that I need to address. > > 1) There was drift during the experiment and there are small translations > between most z-slices > 2) There was one larger jump during the experiment. > > > If I only operate on one channel, StackReg does a good job of removing the > slow drift, but doesn't handle the larger jump (~30-50 pixels). > > I would also like the second channel to be shifted the same as the first > channel, but not separately registered. > > So my two questions are: > 1) Is there a good way to manually shift all the slices after the jump so > StackReg can do its job? > 2) Is there a way to then do the exact same shifts on a parallel stack (or > another channel in a hyperstack)? > > Thanks, > > --David > -- John Kielkopf Professor of Physics and Astronomy University of Louisville Louisville, KY 40292 Tel:502.852.5990 Fax:502.852.0742 |
|
|
A few years ago I modified StackReg to do just what David wants:
http://www.stanford.edu/~bbusse/work/downloads.html Using two simpler alignment steps (one rigid body, the second affine) often handles shifts that a single pass can't. Failing that, if you manually correct the jump in TurboReg and save the transformation, the MultiStackReg jar includes a utility for splicing it into a larger transformation file (Compress Matrices). For more complicated alignment problems such as nonlinear warping, we had the best luck with Fiji's Register/Transform Virtual Stack Slices plugins. Hope it helps! ----- Original Message ----- From: "John Kielkopf" <[hidden email]> To: [hidden email] Sent: Friday, February 24, 2012 11:03:23 AM Subject: Re: Hyperstack registration You could try the Image Stabilizer plugin. It will log translations and apply the same ones again when requested: http://www.cs.cmu.edu/~kangli/code/Image_Stabilizer.html On Fri, Feb 24, 2012 at 10:06 AM, David Hovis <[hidden email]> wrote: > Hello all, > > I have a hyperstack (2-channels, 130 z-slices, 256x256 pixels). I have > two issues with the data that I need to address. > > 1) There was drift during the experiment and there are small translations > between most z-slices > 2) There was one larger jump during the experiment. > > > If I only operate on one channel, StackReg does a good job of removing the > slow drift, but doesn't handle the larger jump (~30-50 pixels). > > I would also like the second channel to be shifted the same as the first > channel, but not separately registered. > > So my two questions are: > 1) Is there a good way to manually shift all the slices after the jump so > StackReg can do its job? > 2) Is there a way to then do the exact same shifts on a parallel stack (or > another channel in a hyperstack)? > > Thanks, > > --David > -- John Kielkopf Professor of Physics and Astronomy University of Louisville Louisville, KY 40292 Tel:502.852.5990 Fax:502.852.0742 |
|
|
In reply to this post by David Hovis
Hi David,
Have you solved your problem? Actually I posted this because I met a problem when proceeded StackReg. I opened my images, and clicked StackReg, but no reaction or processing. I could not find the cause so I wonder maybe you can help me? Sorry for my liberty. Thanks! Tian |
|
|
I believe so. We seem to have settled on SIFT.
My issue was primarily a question of registering images from different frequencies, as opposed to different depths. That seems to work - but I’m not actively involved in it these days. My other stack registration application is more traditional - many slices in Z, all imaged the same way. -- Kenneth Sloan [hidden email] Vision is the art of seeing what is invisible to others. > On Jun 5, 2015, at 15:00 , TTian <[hidden email]> wrote: > > Hi David, > > Have you solved your problem? Actually I posted this because I met a problem > when proceeded StackReg. I opened my images, and clicked StackReg, but no > reaction or processing. I could not find the cause so I wonder maybe you can > help me? Sorry for my liberty. > > Thanks! > Tian > > > > -- > View this message in context: http://imagej.1557.x6.nabble.com/Hyperstack-registration-tp4503024p5013056.html > Sent from the ImageJ mailing list archive at Nabble.com. > > -- > ImageJ mailing list: http://imagej.nih.gov/ij/list.html -- ImageJ mailing list: http://imagej.nih.gov/ij/list.html |
|
|
Hi Kenneth, Thanks for your help. Actually I also tried SPIM , but was always told " no features". Anyway, thanks so much. Regards 2015年6月5日 下午4:05于 "Kenneth R Sloan [via ImageJ]" <[hidden email]>写道：
I believe so. We seem to have settled on SIFT. |
«
Return to ImageJ
|
1 view|%1 views
| Free forum by Nabble | Edit this page |