WinMiPS
Version 1.50
Copyright (c) 1994-95 Christian Buil
____________________________________
INTRODUCTION
============
WinMiPS is a software program especially designed for the
processing of astronomical images from CCD cameras.
WinMiPS contains a set of highly efficient tools which are necessary
to best use your observations. We have placed the emphasis on the
automation of repetitive and fastidious tasks, particularly occurring
during preprocessing operations.
The users of MiPS and QuickMiPS will find it easy to operate
WinMiPS. For instance, many functions have identical parameters
between the DOS and Windows versions. In addition, despite the
graphic interface which characterizes Windows applications, we have
kept in part the line command mode of MiPS. We however did not
preserve the complexities of MiPS such as the scripts, the
redefinition of parameter lists, and arithmetic operation possibilities
found in the command line instructions.
Although the current version of WinMiPS does not have the power
and capabilities of MiPS, for instance, it does not come with the
programming language, we have compensated for this in part by the
high degree of automation of certain functions. Nevertheless, you will
find several powerful features of MiPS in WinMiPS, such as the
entire set of astrometric functions, mosaic, etc... Naturally, WinMiPS
benefits from the Windows user-friendly environment. We are
confident that many users will appreciate this.
Please note that this version of WinMiPS also includes reading
functions of the Hi-SIS22 camera.
This manual only presents a few of the facets of WinMiPS. We have
particularly placed the emphasis on the fundamental image
visualization principles, on the importation of images originating
from various sources and on the preprocessing phases of astronomical
images.
On line help (Help menu) is in fact the reference manual of the
software program. Do not hesitate to refer to it.
SYSTEM CONFIGURATION
====================
Like any image processing software program, and particularly in the
Windows environment, WinMiPS has the minimum system
requirements such as:
*
*
*
*
*
*
Microsoft Windows 3.1 or higher
386 microprocessor
4MB RAM
a hard drive with at least 10MB hard disk space
a video card and a SVGA monitor with 256-color capability
a Microsoft compatible mouse.
WinMiPS performs best with a system of greater capability. The
necessary power is related to the dimension of the images which are
being processed. For instance, a configuration permitting you to
easily work on 512x512 images, requires a 486 microprocessor, 33 MHz
and 8MB RAM.
In order to process 1000x1000 size images, you should use a 486, 50
or 66 MHz, or better a Pentium with 16MB RAM.
The amount of RAM is the most critical issue. If the memory is
insufficient to accept the buffers created by WinMiPS, Windows will
use a virtual memory on the hard drive. This however will result in a
considerable decrease in speed. If you wish to process very large
images, we recommend that you enhance the memory capacity of
your computer (32MB RAM for a 2048x2048 image) rather than the
speed of your microprocessor.
WinMiPS requires a video card with 256-color display, which is the
standard level for all video cards. But for better reproduction,
specifically of three-color images, it is necessary to display 32K, 64K
or 16M colors simultaneously.
LOADING AND SAVING IMAGES
=========================
Before an image can be visualized and processed, it must be
transferred to memory. In fact, in WinMiPS, an image is almost
never completely stored within the Random Access Memory. The
loading operation consists of duplicating the image file in another
file named #0.PIC, called main buffer. This uses a feature already
developed in MiPS. The desired goal is to avoid unnecessarily
overloading the memory.
Then, the visualization and processing will focus exclusively on the
main buffer (unless you indicate otherwise).
Very often, the processing result will be automatically saved in the
main buffer. To avoid loosing this work, should you wish to carry on
additional processing, you must copy the content of the main buffer
in an image file which you will name. This is the saving operation.
In summary, please keep in mind that with WinMiPS, the loading
operation entails duplicating the image of your choice in an image
named #0.PIC. The saving operation is identical.
For reasons which are easy to understand, you should avoid assigning
the name #0.PIC to your images (and more generally a name
beginning with the character # because WinMiPS sometime creates
other buffer files like #1.PIC, #0.BMP,...).
The Load and Save commands are located in the File menu.
We are going to perform the loading of the image M100.PIC (image
obtained with a 190mm diameter telescope, F/D=4, and a Hi-SIS22
camera in composing 6 exposures, each of a 3 minute duration). The
image is located in the WinMiPS installation directory. But before
anything else, we must perform a small operation which will indicate
to WinMiPS where to save the buffer files which it will create during
the processing.
Open the menu Preference and run the command WinMiPS settings. In
the text box Working directory, enter the complete path
of the place of buffer storage. Generally, this directory will contain
the buffers along with the images that you wish to process. Access to
these images will then be facilitated (several functions of WinMiPS
use the working directory as a default directory). In all of the
following examples, the working directory will be: C:\WINMIPS\.
By default, the main buffer file (as well as the secondary buffers) are
not deleted when you quit WinMiPS. This allows you to keep your
work intact at the end of the session. An option of the WinMiPS
Settings command permits you to define whether or not you wish to
erase these files each time you exit from the program.
In using the WinMiPS Settings command, you can define the range
of visualization thresholds between -32768 and 32767 (the image pixels
in the PIC format may be of a negative value). By default, WinMiPS is
set for the visualization of 14 bit images. The modifications made
during your use of WinMiPS are saved in a WINMIPS.INI file when you quit
the program. Thus, you will be able to retrieve your previous settings
in your next work session.
Run the command Load Choose the directory containing the images
(in our examples, we use the directory: C:\WINMIPS), click twice on
the image name. The hard drive will light up for a few seconds: the
image M100.PIC is being copied in #0.PIC.
Please note that the image files, which format originates from
WinMiPS, always uses the extension .PIC. Also note that at this
stage you can change an image in the 8 bit BMP format. This last
image is thus converted to a 16 bit PIC image (#0.PIC) having gray
levels falling between 0 and 255.
IMAGE FILE IMPORTATION AND EXPORTATION
======================================
WinMiPS (as other programs of the MiPS line) uses its own image
format: the PIC format. This format is described in on line help under
the name PIC Format. It possesses a specific heading and offers the
possibility of coding the pixels in several representations (only the
complete 16 bit and real 32 bit formats are recognized in this
program version).
You may load and save images in other formats. To do this you must
utilize the Import and Export commands of the File menu.
The Flexible Image Transport System format (FITS) is commonly
used by professionals worldwide, and is being used with increasing
frequency by amateurs.
An option of the Import command allows automatic loading of an
image in the FITS format (copying this image in the main buffer
#0.PIC). This command automatically recognizes the (complete) 8 bit
and 16 bit FITS formats. Real formats are not accepted.
In the same way, you can convert the image #0.PIC to a FITS format
image (only the 16 bit format is accepted) in using the Export
command.
In choosing the option Free from the Import menu, you have the
possibility of loading images in non-standard formats. You must
specify the width and height of the image in pixels, the coding (8 or
16 bits) and the length in pixels of an eventual heading.
For instance, images from ST4 cameras have a dimension of 192x166
pixels, without a heading and are coded on 8 bits, so:
Dimension X: 192
Dimension Y: 166
Header length: 0
Bits number: 8
An image from a ST6 camera possesses 375x242 pixels, a 2048 pixel
heading and is coded on 16 bits, so:
Dimension X: 375
Dimension Y: 242
Header length: 2048
Bits number: 16
The situation becomes a little more complicated if you do not know
the format of the image. It is then necessary to manually examine the
image header content. For instance, in the case of a ST6 image, you
may type the following command in MS-DOS:
TYPE M31.ST6
Therefore the image heading scrolls just like an ASCII text. The
image width is indicated next to the key word Width and the height
next to the key word Height. The heading dimension remains at 2048
bytes.
You can also export an image in a free format. If you specify that the
heading length is equal to zero, the file that you create contains only
complete 16 bits which describe the image line after line. If you
specify a heading, this one is full of zeros. Please note that you must
indicate the image extension.
The PIC format comprises two comment boxes of 80 characters each.
You may view and edit these boxes using the PIC Header command
from the File menu.
In the same way, you can edit the file heading in FITS format (FITS
Header command). This data is automatically applied to the heading
of the next FITS file that you save.
IMAGE VISUALIZATION
===================
You may proceed in two steps:
* first load the image to be visualized in the main buffer (#0.PIC)
using the command Load... from the File menu.
* then run the command Visu from the View menu, which will
display the main buffer content.
In fact, before being able to display the image, one step remains: the
setting of the visualization threshold.
These thresholds (high and low) define two limits in the image
intensity scale. For visualization, WinMiPS spreads the available
levels of gray (or color in the case of false color visualization)
between these two limits. the effect of the threshold settings is
immediately visible in reduced format in a small window as soon as
the option Continuous is selected.
In reducing the gap between the two thresholds, the contrast is
enhanced.
We have already seen that the maximum limit of the thresholds is
adjustable (command WinMiPS Settings of the Preferences menu).
You may enter thresholds with negative values.
Please note that the fact of having a high threshold inferior to the low
threshold allows for a negative visualization.
Typically, in deep sky imaging, the low threshold is often set in such
a manner that it defines a level of intensity that is just under the
level
of the bottom of the sky (100 to 200 Analog Digital Units). If it is not
the case, the sky will always appear in black which may cause the
loss of certain interesting details.
Click on OK to validate it. The visualization window will then
appear.
While moving the arrow on the image using the mouse, you will
notice that the image coordinates of this arrow will permanently be
displayed on top of the window (the origin is located in the bottom
left corner and has a value of (1,1)) as well as the pixel intensity at
these coordinates.
From the visualization window, you may directly run processing and
analysis commands. These processing operations and analysis always
refer to the displayed image (that is to say the content of the main
buffer).
As an example, we are going to run a function which retrieves the
position of a star's center of gravity. This operation is very useful to
spot the position of stellar objects with precision.
Scroll down the menu located in the upper part of the visualization
window. Choose the option Centroid in clicking on this name (left
button of the mouse).
Then define a rectangular box of 10 to 20 pixels on each side which
includes the star of which you wish to know the position. Slide the
mouse while depressing the right button to do this.
Then, click on the OK button. The coordinates of the star will be
displayed in the exit window.
Now, try the Profile command. In sliding the mouse while
depressing the right button, you define the path of the profile line.
The distance (D) between the two ends of the line and angle (A) in
relation to the horizontal is displayed on top of the window. Click on
OK to display the cut in a specific window.
The duplication of the visualization window is useful for comparison
of images. For this, please choose the command Duplicate and click
on OK. The window which has just been created only permits the
scrolling of the image through the menus. The processing and
analysis operations are only possible in the visualization window.
Among the other available functions from the visualization window
(they are described in the on line help instructions) try to activate the
Neighboring window. Using the mouse, position the end of the
arrow at the level of a dimmer star. Then, while keeping your finger
pressed on the <Ctrl> key, click on the left button of the mouse.
A window displays the pixel intensity located in the area at the end
of the arrow. You may manually modify the intensity of the pixels.
You can also use the Median key which assigning the median
intensity of the group of
the center of the window.
cosmic rays and other hot
changes are automatically
pixels in the window to the pixel located in
This is very useful to manually remove the
areas of the image. In clicking on OK, your
recorded.
There are two images in the upper bar of the visualization window.
The left image is a reduced format representation of the entire
contents of the main buffer. The right image displays the enlarged
box located under the cursor. The command Zoom from the View
menu allows the adjustment of the zoom function.
BASIC PROCESSING
================
Image processing is very simple: with the knowledge of 7 or 8 basic
tools, you can resolve the majority of problems which may arise! The
following operations will illustrate the use of these tools.
Please load the M100.PIC image and view it.
Copy the contents of the visualization window for the purpose of
comparison. Select the Duplicate command from the Window menu.
Then, select the Geometry Tranformation command from the
Processing menu. A window will open offering different geometric
transformations. Select the translation. In the text boxes located in
the bottom, enter the desired translation values in pixels
(DX=horizontal axis, DY=vertical axis). For instance, enter DX=40.3
and DY=22.5, which means that the image is translated towards the
left and top. Please note that you may specify non complete pixel
values (WinMiPS performs the interpolation at a fraction of the
nearest pixel).
The incoming image, in other words, the image to be processed, is
#0.PIC by default. The outcoming image is also #0.PIC, which means
that the contents of the main buffer are going to be modified. You are
perfectly able to specify an outcoming image name different than the
name of the incoming image.
The .PIC extension in the image names is optional. If you do not
specify an extension, WinMiPS will add it automatically.
Please click on OK to execute the command. When this has been
completed, the visualization window will automatically close to
properly indicate that the contents of the main buffer have been
modified (remember that the contents of the last image will be
displayed). This requires you to perform a visualization if you want
to verify the result. Do this and compare it with the starting image.
The translation is a basic tool allowing to superpose several images
in order for example to perform a composite (addition of images).
Although it is not frequently used, try the rotation. Start from the
original M100.PIC image and specify #0.PIC as an outcoming image,
in this way you will be able to immediately view the result. The
program requires the coordinates of the center of rotation (in a
fraction of pixels if you wish). This point may be located outside of
the image (for instance you can assign negative coordinates).
WinMiPS also requires the angle of rotation in degrees. The negative
values are accepted to set the direction of the rotation.
The Mirror options allow for modification of the image orientation
relative to a vertical, horizontal or diagonal pivot.
The Window option permits isolation of a portion of the image. You
must supply two sets of coordinates which define the rectangular box
of the image to be stored.
Finally the Scale option, as its name indicates, performs the image
scale modification. For example, specify a scale factor of 0.5 in X
and in Y to reduce the dimensions of an image by a factor of 2. Scale
factors superior to 1 increase the dimensions of the image. Be careful
not to exaggerate in this area because you can easily end up with
gigantic images. Often the scale increase occurs on a detail of the
image. The Scale command is therefore frequently preceded by the
Window command.
The method used for the change of scale is a simple interpolation of
pixels. The DOS version of MiPS offers techniques of greater
sophistication such as the B-Spline interpolation. Yet you can very
efficiently reduce the pixellisation effect, if you find it undesirable,
in applying a low pass filter to the result (refer to the Gauss option of
the Filtering command of the Processing menu).
In addition to the geometric transformation tools, other functions with
which you should become familiar relate to the arithmetical
operations. Run the command Arithmetic Transformation from the
Processing menu. Numerous arithmetic operations are offered. Some
require two operands, such as the addition or subtraction of two
images. Other only relate to a single image (unique operator) like the
Offset option which consists of adding or removing a constant from
all image pixels.
As a test, subtract the M100.PIC image from itself.
The result is evidently an image where all pixels have a zero value. It
is not always good for the visualization and therefore we add a
constant to the result (here a value of 100).
Perform a visualization with a high threshold of 200 and a low
threshold of 0 to notice that the result is indeed a uniform image.
We are now going to combine the geometric transformations and the
arithmetic transformations. Translate the image M100.PIC of one
pixel to the right and of one pixel to the top and give the name I.PIC
to the result. Then, perform the subtraction of the image M100.PIC
and of the image I.PIC in adding a constant of 500. Visualize the
result with the thresholds (700, 100). You will obtain a nice effect of
directional gradient.
A classic arithmetic operation is to calculate the logarithm of the
image. To obtain an optimal result in the case of deep sky images, it
is necessary to bring the average level of the bottom of the sky
toward the levels 50 to 200. We perform this operation by utilizing
the Offset option and by assigning a value, which is generally
negative, to the offset value.
The level of the bottom of the sky in an image may be estimated in
examining the values of intensity under the mouse arrow when the
arrow is being moved in the image boxes containing no objects. A
more precise method is to statistically analyze the image and to note
the median value of the intensities which is generally a very good
indicator of the level of the bottom of the sky. The statistical analysis
is conducted with the contents of the main buffer by clicking on the
Sum icon located in the upper bar of the main window.
Perform this operation after having loaded the image M100.PIC. You
will obtain a value for the bottom of the sky approaching 400.
Remove the 350 value from all the M100.PIC image pixels and save
the result as I.PIC.
Now calculate the logarithm itself (the value of the norm defines the
maximum amplitude of the final result, here use the value by
default).
At this point you can view the contents of the main buffer by using
the thresholds 3500 and 1500 (high and low thresholds respectively).
Now the spiral structure is clearly visible at the same time as the
faintly lit images boxes.
OPTIMAL REMOVAL OF OBSCURITY SIGNAL
===================================
All CCD images obtained with an integration time greater than a few
seconds are affected by a parasite signal, called the obscurity signal.
The intensity of the signal is linked to the functioning temperature of
the CCD and is directly proportional to the duration of integration. It
is called an obscurity signal because it is completely independent
from the lighting of the detector. Its origin is the thermal disturbance
of the CCD crystalline system.
The removal of the obscurity signal is one of the fundamental steps of
astronomical CCD image preprocessing.
One simple solution to remove the obscurity signal is to follow all
exposures on deep sky objects with an identical exposure performed
in total obscurity. The subtraction of the image from the sky and the
image obtained in obscurity is the solution to our problem.
Although this method is efficient, it also requires a lot of work. In
particular, it necessitates as much time to acquire images from
distant galaxies as to acquire calibration images.
A much more efficient technique is to acquire a single image from
the obscurity signal then to subtract a fraction of this image from the
images to be processed. The whole issue centers on knowing the
multiplication coefficient to be applied to the obscurity reference card
to minimize the thermal noise in the rough images.
This is done through an operation known under the name of Dark
Signal Optimization.
Let's look at an example. Load the image SN.PIC and examine it by
adjusting visualization thresholds. It is a typical obscurity card,
constructed by adding numerous images (a dozen) acquired in long
exposure (several minutes) and in total obscurity. Adding several
elementary images allows for the CCD reading noise to be brought to
a negligible level.
For each obscurity signal card, it is necessary to associate an offset
signal card. The offset signal is obtained in total obscurity with an
allowable exposure as short as possible. The noted offset is related to
the CCD electrical characteristics and to the associated electronics.
The image OF.PIC is a typical offset card created by means of several
elementary images always with the intent of reducing the reading
noise.
It is important to note that the image SN.PIC is the sum of long
exposure images from which the offset card was systematically
removed. Therefore, the image SN.PIC is the true representation of
the obscurity signal.
The analysis of the image SN.PIC shows numerous hot points yet you
will also notice that the average level is close to zero.
With the images SN.PIC and OF.PIC, which can be considered as
constants of the camera, we are going to process the image FL-1.PIC.
Load this last image and view it.
Select the command Dark Signal Optimization from the
visualization window menu. By sliding the mouse with the right
button depressed, define a rectangular box of 40 to 80 pixels on each
side in an area without bright stars. This last point is very important
for an optimal estimation of the multiplication coefficient.
Once you
and will
omit the
the same
have defined the box, click on OK. A window will open
ask for the obscurity card name. Respond with SN (you can
extension). Please note that the obscurity image must be in
directory as the buffers; which is the case here.
You can adapt other parameters by default. Click on OK. WinMiPS
calculates the coefficient value to apply to the obscurity card to
minimize the noise in the image FL-1.PIC, using a dichotomic
method. At the end, the result will appear in the exit window. You
will find a value like 0.19.
We are now going to create our optimal obscurity image to process
FL-1.PIC. Excecute command Arithmetic Transformation from the
Processing menu. Choose the option Mult. by coef. which allows for
multiplying the pixel intensity of an image by a constant. Apply this
processing on the image SN.PIC in selecting the coefficient 0.19.
The operation result is in #0.PIC.
Since we will then have to remove the obscurity signal that we just
calculated as well as the offset signal from the image to be processed,
we will add these two images. Then we will only have one
subtraction to do from this result.
Add the image #0.PIC (the optimal thermal signal) to the image
OF.PIC from the arithmetical operators window. Give the name
N.PIC to the result.
Then you must only subtract the rough image FL-1.PIC from the
obscurity signal card N.PIC (and simultaneously, you will remove the
offset signal). To do this, use the Subtraction option from the
Arithmetic processing window and make sure to assign the constant
a zero value.
Compare the result to the rough image and note the substantial
reduction of noise.
PREPROCESSING
=============
The image N.PIC that you have calculated in the preceding chapter
allows for processing several images which have been acquired in
similar conditions. It is necessary that the images are obtained with
an identical integration time and that there was not too much time
spent between the acquisitions so that the CCD temperature
variations are negligible.
These conditions are respected for the following image sequence: FL1... FL-7. These images will be used later on to calculate a flat-field
card.
To remove the obscurity current from our 7 images, we can use the
arithmetic window 7 times and give a distinct name to the 7 obtained
images, for instance I1.PIC, I2,PIC,... I7.PIC.
It is however tedious and a source of errors. One of the most
powerful features of WinMiPS is to allow for processing in a single
operation of such image sequences.
Execute the command Preprocessing... from the Processing menu.
Select the Subtraction option.
Enter the generic name of the sequence to be processed in the
Generic input box. This name is FL-. WinMiPS will automatically
add the index numbers during processing (1, 2,... 7) as well as the
extension .PIC.
Enter the name of the image to be subtracted, here N, in the box
Second term.
Enter the generic name of the outcoming images, here I. WinMiPS
will automatically add the indices and the extension .PIC, which will
create the images I1.PIC, I2.PIC,..., I7.PIC on your hard drive.
In the box Number (referring to the number of images), type 7.
Assign a zero value to the offset.
Click on OK. The 7 images are automatically processed.
As usual, it is necessary to note that without indication on your part,
the images are searched and saved in the Working directory. This
directory name appears in the upper bar of the window. However you
can instruct WinMiPS to use another directory in specifying a path in
front of the image name.
FLAT-FIELD IMAGE CALCULATION
============================
Flat-field images are used to standardize the response of each image
pixel. The operation consists of dividing the image to be processed
by the flat-field image.
Usually the flat-field image is the average of several images obtained
at dusk, when the stars are not yet visible and when the brightness of
the sky is dark enough to avoid saturating the CCD.
The optimal moment is always quiet brief, thus, it is crucial to be
there at the right time.
It is often easier to extract the flat-field from the observations
conducted during the night or from long exposure images which have
been specifically performed.
The following sequence of images FL-1.PIC,..., FL-7.PIC has been
precisely acquired during the course of the night in order to extract a
flat-field. Please note that each image corresponds to a different field
(this is fundamental). Also note that these exposures have been
obtained with a relatively long integration time in order to get a
significant level of the bottom of the sky. These exposures have also
been obtained without guidance (here this is not critical).
The key to flat-field extraction from stellar images is to eliminate the
stars to preserve the bottom of the sky. This is achieved by using the
median sum
Let's suppose that we process n images. We establish a list of pixel
intensities located at the same X, Y coordinate in each of the images.
This list contains n intensities that we arrange by increasing order.
The median value of the list is the intensity located at the row n/2.
We create a new image in which the pixel of X, Y coordinates is the
median value that we have just calculated. This method allows for
rejecting the statistically less probable information between the two
images.
Based on the approximate pointing of the telescope during the
acquisition, it is less probable that a star will be systematically
aligned on the same pixel within the sequence. The stars are then
naturally deleted during processing.
It is advisable that the images should be preprocessed before
treatment (which implies eliminating the obscurity signal and the
offset signal). It is exactly what we have done during the previous
chapter in creating the sequence I1.PIC, I2.PIC,..., I7.PIC.
A second fundamental precaution to take is that the average level of
images must be identical. If this is not the case, the rejection method
by means of the median will be evidently affected by systematic
effects. Even with exposures of identical duration, the average level
of images may increase from one image to another due to the
brightness variations of the level of the sky (because, for example,
the telescope points lower and lower with respect to the horizon).
Thus, the first task will consist of equalizing the average level of
images of our sequence by multiplying each image by a constant. The
normalization value of the sky background is selected close to the
average level of images (this is not very critical).
The normalization is performed with the option Gain normalization
of the Preprocessing command. from the Processing menu.
In our example, we choose an arbitrary standardization value of 400.
Please note that to save space on a disk we have given the same
generic name to the incoming and outcoming sequence. The
standardized images are therefore: I1.PIC, I2.PIC,..., I7.PIC.
At this point, it is only necessary to perform the median composite in
using the option Median sum in the Preprocessing window.
A statistic appears in the exit window at the end of processing. It
indicates the percentage of each image utilized for producing the
final result. These figures should be comparable (if the percentage is
zero for one or several images, it shows that a problem had arisen in
previous processing or that your starting images do not have
homogeneous characteristics).
Now load the image FLAT.PIC that we have just created and view it
by assigning the value 550 to the high threshold and the value 100 to
the low threshold. The stars have completely disappeared. Only the
bottom of the sky remains with the presence of dust (ring shape spots
originating from the projection of shadows from dust on the CCD).
An optical vigneting effect is also noticeable.
We have just created an excellent flat-field from the images of the
night.
Please note that the median composite is as efficient as there are
many images in the sequence. This type of processing is based on a
statistical analysis which is evidently consolidated in proportion to
the increasing number of measurement possibilities. Usually the
minimum amount of images required to extract a proper flat-field is
7. If your fields are loaded with stars (make sure to avoid the areas of
the milky way) you will need to add a minimum of 11 to 13 images
in order to avoid stellar residue in the result.
Finally please note that because of the nature of extraction of the
median value in the intensity lists, it is necessary that the number of
images within your sequence is odd.
PREPROCESSING OF DEEP-SKY IMAGES
The standard preprocessing of an astronomical CCD deep-sky image
consists of removing the offset level from the rough image, removing
the obscurity current contribution, and finally dividing the resultant
image by the flat-field or ULR in order to standardize the response of
each image pixel.
We are going to process the sequence of rough images M105-1.PIC,
M105-2.PIC and M105-3.PIC. We are equipped with an obscurity
current card SN2.PIC and the flat-field image FL2.PIC.
Here, the offset level of the images to be processed is almost
identical for all the pixels and has a value of 377. Rather than using
the true offset card, we prefer to synthesize an offset image in which
all the pixels are at the 377 level. Therefore, we will not add any
noise during the subtraction of the offset card.
Below is a method to generate this card:
Load the image M105-1.PIC.
Execute the command Image Information from the File menu. Note
the image dimensions (384x256).
Execute the command New Image from the File menu. Give a name
to the image to be synthesized. Here, answer with OF2 which
produces the image OF2.PIC in the working directory. The image
dimensions on the X and Y axis are respectively 384 and 256. All of
the image pixels that you have just produced have an intensity of
zero.
Execute the command Arithmetic Transformation from the
Processing menu. Choose the offset option. The incoming image will
be OF2.PIC and the outcoming image will also be OF2.PIC. The
offset value is 377.
You have just created your offset image OF2.PIC associated with the
images M105.
Apply the dark signal optimization calculation to the image M1051.PIC as explained in the chapter (see above). You should find a
coefficient of approximately 0.31. Multiply the image SN2.PIC by
this constant and remove the result from the image M105-1.PIC.
Also remove the offset OF2.PIC.
At this point, you only need to divide the obtained image by the flatfield image FL2.PIC. For this use the Division option of the
Arithmetic Transformation window. The normalization value to be
supplied is the average intensity of the flat-field image, which here is
1800.
The preprocessing of the image M105-1.PIC is now complete. You
may perform the same operations with the images M105-2.PIC and
M105-3.PIC.
Yet, as we have seen in the chapter PROCESSING, you can
automate these operations in order to simultaneously process the 3
images of the sequence.
But there is a better option. In selecting the option Full processing
from the Preprocessing window, WinMiPS can perform previous
operation all at once. WinMiPS automatically calculates the best
dark signal optimization coefficient for each image, subtracts the
optimal obscurity card, subtracts the offset and divides by the flatfield in searching the standardization coefficient. It is evidently
necessary to provide the generic name of incoming images, the names
of obscurity signal image, of offset and flat-field, as well as the
generic name of the outcoming images. You must also specify the
number of images to be processed as well as the image coordinates of
a rectangular box of 40 to 80 pixels on each side in which there are
no bright stars (or no stars at all if it is possible!).
THE SUMMING
===========
In order to optimize the image processing from the previous chapter,
we need to fuse them together into one single image. It is called the
summing operation.
Due to the fact that the telescope has been slightly moved from one
image to another, it is necessary to superpose these images before
adding them. This is called the registration operation.
In the first image of the sequence (I1.PIC) determine the coordinates
of an isolated star with a precision of 1 or 2 pixels. This star will act
as a reference of position to estimate the relative decentralization on
X and Y between the images of the sequence. You should not choose
a star which is too light so that the program is able to determine its
position with a precision of a fraction of a pixel. For the same reason,
it is absolutely necessary to avoid too bright of stars which give a
saturated image. The star located toward the coordinates (234, 216)
is sufficient for our purposes.
Select the Registration option from the Preprocessing window.
Provide the generic name of the incoming and the outcoming images
(we recommend that the two names are different so as not to ruin the
previous processing work in case of errors). Enter the coordinates of
the star selected in the first image (X and Y parameters). The
parameters Dim. X and Dim. Y correspond respectively to the width
of the box in which the star's center of gravity and the maximum
distance between the images will be calculated.
At this point, it is only necessary to add the images or better yet to
use the median sum or the Sigma clipping technique (see on line
help).
ENHANCEMENT OF THE PLANETARY IMAGE CONTRAST
===========================================
One of the most popular method to enhance the contrast of a
planetary image consists of using the hunsharp masking technique.
Load the image JUPITER.PIC which is a rough image of the planet.
Execute the Filtering command from the Processing menu, then the
Parametric command from the sub-menu. Choose the option
Unsharp Masking with no clipping.
Enter the value 2 for the Sigma parameter and 4 for the Coef.
parameter. Click on OK and compare the obtained image (as usual,
the result is found in the main buffer #0.PIC and, thus, can be viewed
immediately) with the original image.
This type of processing is applicable to stellar images but it is
necessary to select the Unsharp Masking with clipping option,
which imposes a positive constraint on the mask. This avoids the
presence of black aureoles around bright objects of the image (stars).
Let's try another example of planetary image processing. Load the
image JUP1.PIC, e.g. by using the command line. In the Input
window, type:
LOAD JUP1
then press <RETURN> (we suppose that the image is in the working
directory).
Now type:
VISU 4095 1000
The parameters are the high and low visualization thresholds,
respectively.
It is a raw image obtained with a Hi-SIS22 camera in the half-frame
mode (no shutter). The instrument was a 130mm refractor with a
classic two-lens objective and a 2X-barlow lens. The exposure time
was 0.15 s.
We will now use the unsharp masking technique for this image. In
the Input window, type:
UNSHARP
2.5
6
The first parameter is sigma, whereas the second parameter is the
coefficient.
View the result image. Note that you don't need to type the "VISU
4095 1000" text again: just put the cursor anywhere in the
previous visualization command and press <RETURN>.
A very good idea to increase the signal to noise ratio is to add several
images obtained about at the same time in order to neglect the planet
rotation. The series of images JUP1,..., JUP6.PIC corresponds to this
sequence.
Before adding these images, yo need to registrate them. Here is the
method for planetary objects:
Open the Settings window from the visualization window. Modify
the number of pixels used for the centroid computation. In this
example, enter 15000. This corresponds to the number of pixels
covered by the planet.
Then, open the Preprocessing window from the Processing menu
and select the Registration command.
The generic name of the input images is JUP. The generic name of
the output images will be e.g. I. The number of images is 6. Attribute
the coordinates of the center of the image, i.e. 88 & 88 to the X & Y
parameters. Just give the size of the image, i.e. 176, for the size of
the computation window and the search window.
After registrating the images, just add the sequence I1,..., I6 by using
the command Addition from the Preprocessing window. The final
image will be given the name RESULT.PIC.
View the result and then apply the unsharp mask:
LOAD RESULT
VISU 25000 5000
UNSHARP 2.5 6
VISU 25000 5000
NOTE: remember that you may modify the dynamic range of the
thresholds in the WinMIPS Settings window of the Preferences
menu.
DISPLAYING A THREE-COLOR IMAGE
==============================
It is possible to mix the three colored images to obtain an image in
true colors. Use the command Trichromy from the View menu.
Enter the names of the three primary images (respectively the red,
green and blue components) as well as the respective visualization
thresholds. Perform this with the images SATR.PIC, SATV.PIC and
SATB.PIC. For the thresholds choose 3000 for the high threshold and
0 for the low threshold
Please note that you simultaneously create a 24 bit image with a
BMP format.
To benefit from a display in true colors, it is necessary to work with a
graphic mode being able to simultaneously manage 32K, 64K or 16M
colors. The 256 simultaneous color mode is highly insufficient.
THE ASTROMETRIC REDUCTION
=========================
WinMiPS has some extremely precise functions which permit
connecting the X,Y pixel coordinates of objects and their equivalents
in equatorial coordinates.
The astrometric processing of an image is accomplished by the
command Astrometry from the Processing menu.
You must first select several stars within the image of which you
know the equatorial coordinates. You need at least 3 stars.
Today, the availability of CD-ROM databases facilitates this
selection process. The Star Catalog Guide of the Space Telescope
Science Institute should be considered the CD-ROM bible of all
amateurs wanting to precisely determine the position of objects on
these images. Typically, this database permits identifying at least a
dozen stars of which we know the position with a precision of
approximately a second of arc, on a CCD image of about twenty
minutes of arc.
As an example, we are going to astrometrically reduce the image
M100.PIC. 5 stars have been selected. The measurement of the
position of these stars on the image has been obtained with the help
of the command Gaussian PSF from the visualization window menu.
The method consists of applying a mathematical surface to the star
(here a "gaussian") and to note the mathematical center of the
artificial star thus calculated. If the image is followed correctly and
the star sufficiently bright, the precision of the position may be better
than a hundredth of a pixel.
The following are the image coordinates noted for our 5 stars (X
coordinate then Y coordinate):
38.305
186.934
237.553
252.324
349.633
48.573
121.993
184.748
120.466
47.442
The equatorial coordinates of these stars, in degrees and fraction of
degrees, noted in the GSC are respectively (Alpha then Delta):
185.80988
185.70992
185.62463
185.71212
185.81112
15.91689
15.72252
15.65619
15.63681
15.50954
The last parameters to know before being able to do the astronomical
reduction are the approximate coordinates of the center of the image.
Here we have used:
Alpha = 185.62
Delta = 15.66
We now have to create two lists. One containing the Cartesian
coordinates, the other the equatorial coordinates.
The lists are small ASCII files of one or more columns being used as
parameters for certain functions (in MiPS for DOS, the lists are also
frequently used for the automatic photometry reduction commands or
for calculating planetary cards).
Here the lists only include two columns: one for each coordinate (X,
Y or Alpha, Delta). The columns of figures should be separated by at
least one space.
You can create a list outside of WinMiPS by using a text editor
compatible with the ASCII format. You can also use the List
command from the Window menu. Thus you will obtain a small text
editor ideal for this type of application.
The list files must end by the extension .LST. If you use the
WinMiPS editor you can omit this extension during loading and
saving operations because the extension is automatically added as
needed.
Create a list with the Cartesian coordinates that you will call
XY.LST and a list with the equatorial that you will call AD.LST.
Execute the Astrometry command from the Processing menu. At
this time only focus on the Polynomial Calculation box. The
parameter A-D is the name of the equatorial coordinate list. The
parameter X-Y is the name of the Cartesian coordinate list. The Exit
parameter defines the generic name of two polynomial files linking
the two systems of coordinates (one polynomial for each axis). Select
the degree 1 for the polynomials, which here is sufficient. You still
must provide the approximate equatorial coordinates of the center of
the image (in degrees).
Click on Compute. The transfer polynomials between the two
coordinate systems will be calculated.
In the Rectangular to Equatorial section, enter the rectangular
coordinates of a star in the image (X and Y). WinMiPS then will
provide the equatorial coordinates.
The Equatorial to Rectangular section permits you to accomplish
the opposite operation.
From the polynomial file that we have just calculated, it is possible
to create an image which contains a coordinate grid which can easily
supperposed to the image M100.PIC by a simple addition (the grid
image must have the same format as the original image, here
375x253). Use the command Grid from the Processing menu.
ACQUISITION FUNCTIONS
=====================
The version 1.4 of WinMiPS offers acquisition possibilities when
using a Hi-SIS22 camera. To run these commands, the camera must
be connected to the computer and be turned on.
Important: before using the camera, you must verify that the
program is properly configured with your model. In particular, the
command Camera Settings from the Preference menu allows you to
choose between a 12 bit Hi-SIS22 and a 14 bit Hi-SIS22 camera.
Also refer to the appendix of this manual for the parameter setting of
AcqSpeed from the file WINMIPS.INI.
Deep-Sky acquisition
The Deep-Sky window is used for the acquisition of images of long
exposure.
To perform an acquisition, you should select one of the camera
functioning modes (refer to the camera's user manual or on line help),
choose the integration time in seconds and click on OK.
Once the reading of the CCD has been completed, you can use the
Quick-Look button to have a quick look at the result in reduced
format. The visualization thresholds are adjustable (again depress the
Quick-Look button to validate your new choice of thresholds). The
Background button provides the level of the background sky which
will aid in the choice of visualization thresholds.
Please note that you can request the emission of a sound a few
seconds before the end of the exposure. This is useful when you
perform manual obturations.
Planetary acquisition
This Planetary acquisition window allows you to obtain images with
full resolution (pixels of 9x9 microns) having a format of 176x176
pixels.
The window can be located in the center of the CCD or in the top
part to initiate the half frame acquisition technique (we will use this
acquisition mode in planetary imaging if we do not have an
obturator).
The obtained image will be displayed in the left box. You may save
this image for comparison by copying it in the right box. Click on the
arrow located between the two visualization boxes.
The acquisitions may be completed by either allowing for a stop
between each image or in continuous mode.
You can save the last acquired image in assigning it any name (never
specify the extension .PIC). WinMiPS always adds the index value at
the end of this name if it is stated.
It is possible to increase the index by one unit each time you save an
image. This is very convenient to automatically save the image
sequences (for instance: JUP1.PIC, JUP2.PIC, JUP3.PIC...).
Focusing
Here, WinMiPS only displays the profile of an image line. The line
number can be fixed (Fixed mode) or determined by the program in
such a manner as to pass by the brightest point of the image (Track
mode). This acquisition mode is used to focus on stellar objects.
APPENDIX 1: the file WINMIPS.INI
================================
The file WINMIPS.INI contains several variables related to
acquisition with Hi-SIS22 cameras. This file is automatically created
in the WINDOWS directory after the first session of work with
WinMiPS.
Variable: CameraType
The contents of this variable determine the type of camera to be
used. In this WinMiPS version the accepted values are 2 for a 12
bit Hi-SIS22 and 3 for a 14 bit Hi-SIS22. Default: 14 bits camera.
Variable: AcqSpeed
The first Hi-SIS22 cameras were equipped with a microprocessor
functioning at a 12MHz frequency. The current models (cameras
available as of February 1994) function at a 16MHz frequency.
The AcqSpeed must take the value 1105.92 for the old models and
1600 for the new models.
VERY IMPORTANT: The AcqSpeed variable is set at 1600 by
default. It is crucial to modify its contents with a text editor
(EDIT command of MS-DOS 6.0 for example) if you use an old
model camera.
Please note that it is possible to upgrade a 12MHz camera to a
16MHz camera. Contact dealer for details.
Variable: Port
Select printer port: 1=LPT1, 2=LPT2. By default: LPT1.
Variables: Time1, Time2, Time3, Time4, Time5
The contents of these times refer to temporization during the
reading of the CCD. By default these variables contain the value 5.
It is possible to accelerate the reading time of the CCD by reducing
these values. You must however perform a few successive tests in
changing only one variable at a time to make sure that the camera
continues to function properly (absence of parasites, low noise
level,...) The minimum value of these variables is 1. They can be
adjusted within WinMiPS (command Camera Settings from the
Preferences menu). The following is a typical configuration for a
PC486-66MHz:
TIME1=2
TIME2=3
TIME3=2
TIME4=3
TIME5=2
Variable: Country
Country=US for USA, Country=FR for France.
APPENDIX 2: images
==================
N4565.PIC: Lichtenecker flat-field camera 190 mm Fnumber=4. Hi-SIS22,
binning 2x2.
M63.PIC: T60 Pic du Midi Observatory (24 inch Fnumber=3.5). Hi-SIS22,
binning 2x2 (Association T60).
M51-1.PIC: T60 Pic du Midi Observatory (24 inch Fnumber=3.5). Hi-SIS22,
binning 2x2 (Association T60).
M51-2.PIC: Epsilon 160 Takahashi (Fnumber=3.3). Hi-SIS22, binning 1x1.
JPU1..5.PIC: 5 inch refractor Fnumber=15 + Barlow 2x.
Hi-SIS22 half-frame mode.
MOON.PIC : Mewlon 210 Takahashi. Hi-SIS22.