Plugin announcement: MosaicJ

[Philippe Thévenaz]

###
#
# MosaicJ
#
# Version dated May 23, 2007
#
###

Here is a new plugin that may be useful to those of you who want to  
stitch together partial images so as to create a larger mosaic image.
The purpose of this plugin is to facilitate the assembly of a mosaic  
of overlapping individual images, or tiles. It provides a semi-
automated solution where the initial rough positioning of the tiles  
must be performed by the user, and where the final delicate  
adjustments are performed by the plugin. The MosaicJ plugin requires  
that a second plugin, named TurboReg, is installed.

MosaicJ and its documentation is available from here:
http://bigwww.epfl.ch/thevenaz/mosaicj/

TurboReg and its documentation is available from here:
http://bigwww.epfl.ch/thevenaz/turboreg/

===
EPFL/STI/IOA/LIB/BM.4.137
P. Thévenaz
Station 17
CH-1015 Lausanne VD
===
phone (CET): +41(21)693.51.61
fax: [phone]
RFC-822: [hidden email]
X-400: /C=ch/A=400net/P=switch/O=epfl/S=thevenaz/G=philippe/
URL: http://bigwww.epfl.ch/
===

[Malgorzata Swider]

I am testing this plug-in and I got stuck right at the beginning. I loaded 4
square images for my mosaic and they appeared in the lower area of the
screen, so far so good. When I tried to activate one of them by clicking on
it, it was moved to the upper area of the screen but appeared as a
completely black square. What should I do?
Thanks,
MS

----- Original Message -----
From: "Philippe Thévenaz" <[hidden email]>
To: <[hidden email]>
Sent: Thursday, June 07, 2007 3:36 AM
Subject: Plugin announcement: MosaicJ

> ###
> #
> # MosaicJ
> #
> # Version dated May 23, 2007
> #
> ###
>
> Here is a new plugin that may be useful to those of you who want to
> stitch together partial images so as to create a larger mosaic image.
> The purpose of this plugin is to facilitate the assembly of a mosaic  of
> overlapping individual images, or tiles. It provides a semi- automated
> solution where the initial rough positioning of the tiles  must be
> performed by the user, and where the final delicate  adjustments are
> performed by the plugin. The MosaicJ plugin requires  that a second
> plugin, named TurboReg, is installed.
>
> MosaicJ and its documentation is available from here:
> http://bigwww.epfl.ch/thevenaz/mosaicj/
>
> TurboReg and its documentation is available from here:
> http://bigwww.epfl.ch/thevenaz/turboreg/
>
> ===
> EPFL/STI/IOA/LIB/BM.4.137
> P. Thévenaz
> Station 17
> CH-1015 Lausanne VD
> ===
> phone (CET): +41(21)693.51.61
> fax: [phone]
> RFC-822: [hidden email]
> X-400: /C=ch/A=400net/P=switch/O=epfl/S=thevenaz/G=philippe/
> URL: http://bigwww.epfl.ch/
> ===

[Albert Cardona]

Philippe,

TrakEM2 provides a free canvas where images can float photoshop-style.
The advantage should be obvious: when algorithms fail, images can be
adjusted manually simply by dragging them. In addition, single images
can be automatically stitched to others by dragging and snapping (see
http://www.ini.uzh.ch/~acardona/trakem2_manual.html#semi_snap )

It would be great if MosaicJ was written in such a manner that one could
hand it a list of images and their approximate coordinates, and retrieve
from it the transformation matrices for each image.

At the moment TrakEM provides the means to import montages where tile
position is roughly known, and finely stitch tiles with
cross-correlation. See examples here:

http://www.ini.uzh.ch/~acardona/img/grid-montage-cross-correlation.jpg
http://www.mcdb.ucla.edu/research/hartenstein/acardona/viewer.html?img/first_instar_brain_lobe.jpg&First%20instar%20brain%20lobe%20montage

Some details on how it works here:

http://www.ini.uzh.ch/~acardona/trakem2_manual.html#importing_grids
http://www.ini.uzh.ch/~acardona/trakem2_manual.html#importing_seq_grids


Get latest development version of TrakEM2 here:
http://www.ini.uzh.ch/~acardona/edge/TrakEM2_.jar
http://www.ini.uzh.ch/~acardona/edge/TrakEM2-src.zip
http://www.ini.uzh.ch/~acardona/edge/ImageJ_3D_Viewer.jar

(Java 1.6.0, and requires ImageJ_3D_Viewer.jar from Bene Schmid linked
above)

TrakEM2 is changing a lot as of this moment, we are on a hackathon
working on ImageJ-related projects.

Albert

[Pascal Lorentz]

Dear all,

This also happend to me. Using this plugin on a PC leads to the described black
squares. I also tried it on a MAC and there no problems occured.
But I would really like to use it on my PC as well.

Cheers

Pascal

Zitat von Malgorzata Swider <[hidden email]>:

> I am testing this plug-in and I got stuck right at the beginning. I loaded 4
> square images for my mosaic and they appeared in the lower area of the
> screen, so far so good. When I tried to activate one of them by clicking on
> it, it was moved to the upper area of the screen but appeared as a
> completely black square. What should I do?
> Thanks,
> MS
>
> ----- Original Message -----
> From: "Philippe Thévenaz" <[hidden email]>
> To: <[hidden email]>
> Sent: Thursday, June 07, 2007 3:36 AM
> Subject: Plugin announcement: MosaicJ
>
>
> > ###
> > #
> > # MosaicJ
> > #
> > # Version dated May 23, 2007
> > #
> > ###
> >
> > Here is a new plugin that may be useful to those of you who want to
> > stitch together partial images so as to create a larger mosaic image.
> > The purpose of this plugin is to facilitate the assembly of a mosaic  of
> > overlapping individual images, or tiles. It provides a semi- automated
> > solution where the initial rough positioning of the tiles  must be
> > performed by the user, and where the final delicate  adjustments are
> > performed by the plugin. The MosaicJ plugin requires  that a second
> > plugin, named TurboReg, is installed.
> >
> > MosaicJ and its documentation is available from here:
> > http://bigwww.epfl.ch/thevenaz/mosaicj/
> >
> > TurboReg and its documentation is available from here:
> > http://bigwww.epfl.ch/thevenaz/turboreg/
> >
> > ===
> > EPFL/STI/IOA/LIB/BM.4.137
> > P. Thévenaz
> > Station 17
> > CH-1015 Lausanne VD
> > ===
> > phone (CET): +41(21)693.51.61
> > fax: [phone]
> > RFC-822: [hidden email]
> > X-400: /C=ch/A=400net/P=switch/O=epfl/S=thevenaz/G=philippe/
> > URL: http://bigwww.epfl.ch/
> > ===
>

--
Pascal Lorentz
Centre for Biomedicine
Department of Clinical-Biological Sciences
University of Basel
Mattenstrasse 28
4058 Basel
Switzerland

[hidden email]
www.cbm.unibas.ch

----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.

[Gabriel Landini]

This plugin works very well.
I wonder if it is possible to induce some automated alignment of the current
image versus the overlap, while in the panel, so if the alignment fails one
can try to resposition the image?

Regards,

Gabriel

[Jan Eglinger-5]

Dear all,

Importing gray scale as well as RGB tif images worked fine for me on PC
(working with ImageJ v1.38p on Windows XP), no black squares and easy
drag'n'drop alignment,

however, when executing "Create mosaic", nothing seems to happen at all.
The log file is created but remains at 0kb size (no matter how long I
wait, I haven't tried waiting hours though ;-)

Cheers,

jan

-------- Original Message  --------
Subject: Re:Plugin announcement: MosaicJ
From: Pascal Lorentz <[hidden email]>
To: [hidden email]
Date: 07.06.2007 17:32

> Dear all,
>
> This also happend to me. Using this plugin on a PC leads to the described black
> squares. I also tried it on a MAC and there no problems occured.
> But I would really like to use it on my PC as well.
>
> Cheers
>
> Pascal
>
> Zitat von Malgorzata Swider <[hidden email]>:
>
>> I am testing this plug-in and I got stuck right at the beginning. I loaded 4
>> square images for my mosaic and they appeared in the lower area of the
>> screen, so far so good. When I tried to activate one of them by clicking on
>> it, it was moved to the upper area of the screen but appeared as a
>> completely black square. What should I do?
>> Thanks,
>> MS
>>
>> ----- Original Message -----
>> From: "Philippe Thévenaz" <[hidden email]>
>> To: <[hidden email]>
>> Sent: Thursday, June 07, 2007 3:36 AM
>> Subject: Plugin announcement: MosaicJ
>>
>>
>>> ###
>>> #
>>> # MosaicJ
>>> #
>>> # Version dated May 23, 2007
>>> #
>>> ###
>>>
>>> Here is a new plugin that may be useful to those of you who want to
>>> stitch together partial images so as to create a larger mosaic image.
>>> The purpose of this plugin is to facilitate the assembly of a mosaic  of
>>> overlapping individual images, or tiles. It provides a semi- automated
>>> solution where the initial rough positioning of the tiles  must be
>>> performed by the user, and where the final delicate  adjustments are
>>> performed by the plugin. The MosaicJ plugin requires  that a second
>>> plugin, named TurboReg, is installed.
>>>
>>> MosaicJ and its documentation is available from here:
>>> http://bigwww.epfl.ch/thevenaz/mosaicj/
>>>
>>> TurboReg and its documentation is available from here:
>>> http://bigwww.epfl.ch/thevenaz/turboreg/
>>>
>>> ===
>>> EPFL/STI/IOA/LIB/BM.4.137
>>> P. Thévenaz
>>> Station 17
>>> CH-1015 Lausanne VD
>>> ===
>>> phone (CET): +41(21)693.51.61
>>> fax: [phone]
>>> RFC-822: [hidden email]
>>> X-400: /C=ch/A=400net/P=switch/O=epfl/S=thevenaz/G=philippe/
>>> URL: http://bigwww.epfl.ch/
>>> ===
>
>
> --
> Pascal Lorentz
> Centre for Biomedicine
> Department of Clinical-Biological Sciences
> University of Basel
> Mattenstrasse 28
> 4058 Basel
> Switzerland
>
> [hidden email]
> www.cbm.unibas.ch
>
> ----------------------------------------------------------------
> This message was sent using IMP, the Internet Messaging Program.

--
Jan Eglinger
PhD Student, Eckhard Lammert group
Max Planck Institute of Molecular Cell Biology and Genetics
Pfotenhauerstrasse 108
01307 Dresden
Germany

Tel:  [phone]
VoIP: [hidden email]

[Gabriel Landini]

On Thursday 07 June 2007, Jan Eglinger wrote:
> however, when executing "Create mosaic", nothing seems to happen at all.

Not sure if this will help, but something similar happened to me, until I
realised that the version of Turbo_Reg I had was too old.

Cheers,

G.

[Philippe Thévenaz]

> From:  [hidden email]
> Subject: Re: Plugin announcement: MosaicJ
> Date: 7 juin 2007 18:09:05 GMT+02:00
> To:  [hidden email]
> Reply-To:  [hidden email]
>
> This plugin works very well.
> I wonder if it is possible to induce some automated alignment of  
> the current
> image versus the overlap, while in the panel, so if the alignment  
> fails one
> can try to resposition the image?
>
> Regards,
>
> Gabriel

That's not so easy. But something not too remote is already possible:
1) Prepare your coarse mosaic;
2) Create the final mosaic [File -> Create Mosaic];
3) Suppose this fails (sorry for the inconvenience). Instead of  
despairing, do
4) Relaunch MosaicJ;
5) Restore your coarse mosaic [File -> Previous Pre-Mosaic];
6) Adjust whatever you believe is worth adjusting;
7) Move on with step 2) and cross fingers that it works better this  
time.

[Philippe Thévenaz]

> From:  [hidden email]
> Subject: Re: Plugin announcement: MosaicJ
> Date: 7 juin 2007 16:30:34 GMT+02:00
> To:  [hidden email]
> Reply-To:  [hidden email]
>
> Philippe,
>
> TrakEM2 provides a free canvas where images can float photoshop-style.
> The advantage should be obvious: when algorithms fail, images can  
> be adjusted manually simply by dragging them. In addition, single  
> images can be automatically stitched to others by dragging and  
> snapping (see http://www.ini.uzh.ch/~acardona/ 
> trakem2_manual.html#semi_snap )
>
> It would be great if MosaicJ was written in such a manner that one  
> could hand it a list of images and their approximate coordinates,  
> and retrieve from it the transformation matrices for each image.
>
> [...]
>
> Albert

MosaicJ is indeed written in such a manner that one can hand it a  
list of images and their approximate coordinates, and retrieve from  
it a description of the transformation for each image.

The list of images and their approximate coordinates can be given in  
a pre-mosaic file.

The transformations can be retrieved from the log file, which is  
created on-demand.

[Philippe Thévenaz]

I don't have access to a wide range of PCs, so I cannot test a lot of  
configurations. On the (one, old) machine I have access to, I did not  
encounter the problem you describe. I was using the version of ImageJ  
for Windows that is distributed with a bundled Java Virtual Machine.  
Perhaps that makes a difference?

http://rsb.info.nih.gov/ij/download/win32/ij137-jdk15-setup.exe

===
EPFL/STI/IOA/LIB/BM.4.137
P. Thévenaz
Station 17
CH-1015 Lausanne VD
===
phone (CET): +41(21)693.51.61
fax: [phone]
RFC-822: [hidden email]
X-400: /C=ch/A=400net/P=switch/O=epfl/S=thevenaz/G=philippe/
URL: http://bigwww.epfl.ch/
===

> From:  [hidden email]
> Subject: Re: Plugin announcement: MosaicJ
> Date: 7 juin 2007 17:32:09 GMT+02:00
> To:  [hidden email]
> Reply-To:  [hidden email]
>
> Dear all,
>
> This also happend to me. Using this plugin on a PC leads to the  
> described black
> squares. I also tried it on a MAC and there no problems occured.
> But I would really like to use it on my PC as well.
>
> Cheers
>
> Pascal
>
> Zitat von Malgorzata Swider <[hidden email]>:
>
>
>> I am testing this plug-in and I got stuck right at the beginning.  
>> I loaded 4
>> square images for my mosaic and they appeared in the lower area of  
>> the
>> screen, so far so good. When I tried to activate one of them by  
>> clicking on
>> it, it was moved to the upper area of the screen but appeared as a
>> completely black square. What should I do?
>> Thanks,
>> MS
>>
>> ----- Original Message -----
>> From: "Philippe Thévenaz" <[hidden email]>
>> To: <[hidden email]>
>> Sent: Thursday, June 07, 2007 3:36 AM
>> Subject: Plugin announcement: MosaicJ
>>
>>
>>
>>> ###
>>> #
>>> # MosaicJ
>>> #
>>> # Version dated May 23, 2007
>>> #
>>> ###
>>>
>>> Here is a new plugin that may be useful to those of you who want to
>>> stitch together partial images so as to create a larger mosaic  
>>> image.
>>> The purpose of this plugin is to facilitate the assembly of a  
>>> mosaic  of
>>> overlapping individual images, or tiles. It provides a semi-  
>>> automated
>>> solution where the initial rough positioning of the tiles  must be
>>> performed by the user, and where the final delicate  adjustments are
>>> performed by the plugin. The MosaicJ plugin requires  that a second
>>> plugin, named TurboReg, is installed.
>>>
>>> MosaicJ and its documentation is available from here:
>>> http://bigwww.epfl.ch/thevenaz/mosaicj/
>>>
>>> TurboReg and its documentation is available from here:
>>> http://bigwww.epfl.ch/thevenaz/turboreg/
>>>
>>> ===
>>> EPFL/STI/IOA/LIB/BM.4.137
>>> P. Thévenaz
>>> Station 17
>>> CH-1015 Lausanne VD
>>> ===
>>> phone (CET): +41(21)693.51.61
>>> fax: [phone]
>>> RFC-822: [hidden email]
>>> X-400: /C=ch/A=400net/P=switch/O=epfl/S=thevenaz/G=philippe/
>>> URL: http://bigwww.epfl.ch/
>>> ===
>>>
>>
>>
>
>
> --
> Pascal Lorentz
> Centre for Biomedicine
> Department of Clinical-Biological Sciences
> University of Basel
> Mattenstrasse 28
> 4058 Basel
> Switzerland
>
> [hidden email]
> www.cbm.unibas.ch
>
> ----------------------------------------------------------------
> This message was sent using IMP, the Internet Messaging Program.

[M. Nur Heriawan]

Dear ImageJ`ers,

Firstly, I would like to make sure...is it correct
that the CT values resulted from X-ray CT scanner is
only ranging from -1000 to 2000? So, if we consider
the relationship between CT value and bulk density of
material (rho) is:

CT value = 1000*rho - 1000

then only material with bulk density from 0 to
3 g/cc can be detected by using X-ray CT scanning.

In case our material contains a small material with
bulk density more than 3 g/cc (for example: mineral
pyrite has density around 5 g/cc), how we can detect
it from the CT image?

Secondly, after I did thresholding using ImageJ for my
CT image, I found that it contains of `cupping effect`
with darker color presents around the center part of
image. Cupping effect is one of the artificial effect
on CT image due to the beam hardening effect during
scanning process. Does anyone have experience to
remove this cupping effect by image processing?

Any comments and suggestions are very welcome.

Thank you.


---------------
M. Nur Heriawan
Laboratory of Applied Geoscience and Technology  
Graduate School of Science and Technology
Kumamoto University, JAPAN
http://www.civil.kumamoto-u.ac.jp/tansa


       
____________________________________________________________________________________
Choose the right car based on your needs.  Check out Yahoo! Autos new Car Finder tool.
http://autos.yahoo.com/carfinder/

[Harry Parker]

Hello, M. Nur Heriawan.

I do not know much about CT imaging, but I am a physicist and a photographer.

First, the relation between CT values and rho is dependent on many factors such as the exposure level and energy of the X-rays in the beam. So we cannot tell what is the correct equation. That depends on your setup. If the material density is too high for your current settings, the exposure needs to be increased one way or another.  Perhaps different exposures will be required to get good data for different parts of the sample. This is one way to increase the dynamic range of the data.


Perhaps I can help with your second problem, the "cupping" effect. Regular camera images also have similar nonuniformities due to "lens shading" effects, sometimes called "vignetting".

In digital photography, images do not have negative values. What do
they indicate in CT?  The "zero" /"no signal" value needs to be determined for your system. The constant offset must be subtracted out. This technique also assumes that the signal level is linear to exposure. So if your recorded values are nonlinear (e.g. logarithmic) this needs to be corrected as well.


Basically, less light reaches the sides and corners of an image than reaches the center.  So an image of a uniform scene will not be uniform, but darker in the corners.

This can be corrected by multiplying each pixel by a factor that is the inverse of the falloff.  This factor is 1.0 near the center of the image ( = no correction) and increases towards the edges and corners.

This effect can be measured and eliminated by taking images of a blank uniform scene. Getting a really uniform scene is the hardest part of this technique. Many blank images should be averaged together to minimize noise. Then convert the image to 32 bit (floating point) if it is not already in that format.

Then divide the whole image by its center intensity. Now you have a "gain image" where each pixel's value is its relative gain coefficient. Use the Image Calculator function to divide your object images by this gain image to correct them.

If the falloff is large in the corners, then the gain will be great in the corners, greatly increasing the noise in that part of the image.  This limits the practical value of this technique.

Does this help?
 
--  
Harry Parker  
Senior Systems Engineer  
Digital Imaging Systems, Inc.

----- Original Message ----
From: M. Nur Heriawan <[hidden email]>
To: [hidden email]
Sent: Friday, June 8, 2007 7:21:37 AM
Subject: CT value and cupping effect

Dear ImageJ`ers,

Firstly, I would like to make sure...is it correct
that the CT values resulted from X-ray CT scanner is
only ranging from -1000 to 2000? So, if we consider
the relationship between CT value and bulk density of
material (rho) is:

CT value = 1000*rho - 1000

then only material with bulk density from 0 to
3 g/cc can be detected by using X-ray CT scanning.

In case our material contains a small material with
bulk density more than 3 g/cc (for example: mineral
pyrite has density around 5 g/cc), how we can detect
it from the CT image?

Secondly, after I did thresholding using ImageJ for my
CT image, I found that it contains of `cupping effect`
with darker color presents around the center part of
image. Cupping effect is one of the artificial effect
on CT image due to the beam hardening effect during
scanning process. Does anyone have experience to
remove this cupping effect by image processing?

Any comments and suggestions are very welcome.

Thank you.


---------------
M. Nur Heriawan
Laboratory of Applied Geoscience and Technology  
Graduate School of Science and Technology
Kumamoto University, JAPAN
http://www.civil.kumamoto-u.ac.jp/tansa


       
____________________________________________________________________________________
Choose the right car based on your needs.  Check out Yahoo! Autos new Car Finder tool.
http://autos.yahoo.com/carfinder/





       
____________________________________________________________________________________
Moody friends. Drama queens. Your life? Nope! - their life, your story. Play Sims Stories at Yahoo! Games.
http://sims.yahoo.com/ 

[Francesco Lassandro]

----- Original Message -----
From: "M. Nur Heriawan" <[hidden email]>
To: <[hidden email]>
Sent: Friday, June 08, 2007 1:21 PM
Subject: CT value and cupping effect


> Dear ImageJ`ers,
>
> Firstly, I would like to make sure...is it correct
> that the CT values resulted from X-ray CT scanner is
> only ranging from -1000 to 2000?

Not exactly. the available range of CT numbers varies with scanners and bits
per pixel.
In this scale the unit is the Hounsfield unit (HU), -1000 represents the
attenuation of air, 0 is the attenuation of air. There is no upper limit to
the scale.


So, if we consider

> the relationship between CT value and bulk density of
> material (rho) is:
>
> CT value = 1000*rho - 1000
>
> then only material with bulk density from 0 to
> 3 g/cc can be detected by using X-ray CT scanning.
>
> In case our material contains a small material with
> bulk density more than 3 g/cc (for example: mineral
> pyrite has density around 5 g/cc), how we can detect
> it from the CT image?

I think you can't with conventional machines. furthermore, the relationship
between rho and CT value
holds true when atomic numbers of the substance is similar to that of water.
Anyway attenuation of substances with a high atomic number is strongly
influenced by the energy of the x-ray beam. With a 12-bit encoding of
CT-numbers are outside the range. Metals may cause almost complete
absorption with artifacts. 16-bit encoding scanners (maximum CT-number
>65,000 UH) may permit to evaluate also metallic structure.

ciao
Francesco