| Description: |
The program patchwork_ (running under ImageJ version
1.27z) is able to combine images that are overlapping (mosaic). The
program was tested on bright transmission images coming from a
microscope. With this program, the user can combine all the images
of a mosaic into one. The program can handle 8bit, 16bit and color
(RGB) images. The option of correcting the flaws of the images is
also given (See Background Correction). The images should be aligned
(which means that you don't have to shift them in 2 dimension to
find the match, but only in one).
A Few Guidelines
The main program is the "patchwork_" one. All the other classes are
used by this one. Class patchwork_ uses classes: ...correl2_
...correl3_ ...Final_proc_ ...File Opener_ (downloaded from ImageJ
PlugIn list)
1. Compile and Run patchwork_ .java
2. The user should provide the number of the images he/she wants to
combine, in the beginning.
3. The user should open the images in the correct order (from the
one in the right bottom to the one on the left top, moving upwards.
Imagine the mosaic as a matrix of images and start from element
(n,m), then open (n-1,m)...,(n,m-1),(n-1,m-1)...to (0,0). The File
Opener plugin is used (as found on the webpage) in order to open the
images. The user can also open the images before running the program
and choose when the File Opener runs.
4. The user should provide the number of the "rows" of the mosaic,
the percentage of overlap vertically and the percentage of overlap
horizontally.
5. The option of eliminating background abnormalities is given. The
images are divided by a "background image". Click OK to implement
Background correction (for better results). If you don't want
background correction, just change the default option and click OK.
When the user opens the background image, the program starts
combining the images. 6. Let it run until you are asked to save the
final image or until an error message about the input appears.
7. Save the final image otherwise you lose it!
Installation and Documentation
In order to run the patchwork_
program to turn a mosaic into a single image, download the .java
files of the classes involved (patchwork_.java, correl2_.java,
correl3.java, Final_proc.java, File Opener.java). Compile each one
of these files, ignoring the NullPointerException and canceling the
File Opener_ run. Compile patchwork_ in the end and run it. The
comments in these programs are written in such a way as to be easy
to create a documentation file (HTML) by using Doxygen program (run
the Doxygen wizard and use the above .java files as input).
(download it from:
http://www.stack.nl/~dimitri/doxygen/download.html#latestsrc)
Detailed explanation of the purpose of patchwork_ program
When we want to examine an object (e.g. cell) with a microscope, we
prefer to zoom in in certain areas of this object to get better
resolution, but we also want to have the whole picture. So we create
a mosaic by capturing parts of the object and getting images that
overlap (we need the overlap to reconstruct the object again).
Reconstructing the whole image from some smaller ones is the purpose
of this program.
Background Correction
When capturing bright transmission images the
result is not always satisfying. For example, there may be stains on
the lenses and, usually, the lighting is not homogenous all over the
image. These may cause problems when we try to reconstruct the whole
picture. In this program, there been an effort to correct this by
taking an image with no object inside (with the microscope), which
unavoidably has all the flaws described above. We divide this image
by every other image that is part of the mosaic and we get corrected
images to combine.
Example
If you want to view a little demonstration of this program,
you can use the images that are in the following address.
www.photos.yahoo.com/kavagiouzaharoula (Image folder) Download
them and save them in the order they are numbered so that you can
open them easily when you run the program. The mosaic is created by
4 8bit grayscale images and you can use the "background image" to
correct the other 4. The correct answers for the program are:
number of images -> 4
number of rows -> 2
vertical overlap -> 40 (%)
horizontal overlap -> 27 (%)
|