Previous Message | Next Message Previous in Topic | Next in Topic Previous by Same Author | Next by Same Author Previous Page (May 2002) | Back to Main IMAGEJ Page
	Reply | Post a New Message | Join or Leave IMAGEJ, or Change Options | Search
	Chronologically | Most Recent First | Wrap Text (Proportional Font) | Don't Wrap Text (Non-proportional Font)

Content-Type: text/plain; charset="iso-8859-1"
Message-ID:  <[log in to unmask]>
Date:         Tue, 28 May 2002 08:51:18 -0500
Reply-To:     ImageJ Interest Group <[log in to unmask]>
Sender:       ImageJ Interest Group <[log in to unmask]>
From:         Rex Couture <[log in to unmask]>
Subject:      Re: Image registration and alignment

At 06:09 PM 5/27/02 +0200, Tom Andersen wrote:
>I'm trying to register two images using the TurboReg plug-in. As far as I can tell,
>the plug-in always returns a 32-bit gray (!) stack of two images.

        That's a pretty good program, but there seem to be a lot of errors, and quite a few practical problems with using it.  I don't have time to look at it or fix it, but maybe I can offer some suggestions for Dr. Thévenaz, point out some of the problems, and answer some of Tom's questions.

        Sometimes if I start with 16-bit unsigned integer images, the 32-bit result is correct, sometimes not.  Sometimes the result values are large negative (e.g. -60000) or impossibly large positive values (e.g. 67000).  It seems to work if I convert the images to 32 bits first.

>The first image
>in the resulting stack is the transformed source image, and the second is a mask
>layer, which, on my machines always is white (does anyone know the function of this
>mask layer?). The image layer of the stack has higher contrast than the initial
>source image.

        If you read the documentation carefully, you will find that the mask limits the area that is used in registration.  You can add a stack to your source image, select one or more regions of interest, switch to the second image in the stack, and fill the ROI with a nonzero value.  In your case, the entire image is used for registration.

****************************
        Suggestion to Dr. Thévenaz:  It would be extremely helpful if the target image mask were used instead of the source image mask.  In that way, multiple images could be registered to the same image, using the same mask.  The way it the program is written now, the mask must be reconstructed for each image.  This is time-consuming, and the mask will always be different.
****************************

>Is there any way to: ...
>    - make the cross-hairs for the adjustment points change with the image
>resolution?

        You can magnify the image while you are running the plugin.  This helps somewhat.


Med venlig hilsen,

Rex Couture, Ph.D.
Electronic Radiology Laboratory
Dept. of Radiology
Washington University School of Medicine
Campus Box 8225
510 S. Kingshighway
St. Louis MO 63110
(314) 747 0332

[log in to unmask]

Center for Information Technology National Institutes of Health Bethesda, Maryland 20892 301 594 6248 (v) 301 496 8294 (TDD) Comments and Assistance Accessibility