ACQUISITION OF IMAGES USED FOR PRE-PROCESSING

In deep sky astrophotography, it is necessary to acquire images used to correct for parasites, such as thermal noise, coming from the camera when taking long exposure photos. Also these images can correct vignetting problems due to the optics used. These images will be removed from the images of the deep sky object to eliminate these defects in the resulting image. There are three types of images to acquire:

1. The preload image (Offset or Bias)
2. The Black (Dark)
3. The uniform light range (PLU or flat field)

To acquire the preprocessing images, it is advisable to use the same software that will be used for the acquisition and preprocessing of the deep sky images. The results will thus be maximized. Most of the time, the file format is FIT. To take the preprocessing images, the software takes the images in black and white RAW mode for color CCD cameras (some digital cameras also provide images in RAW mode). Even if the image appears in black and white, this mode preserves the colors. Regarding the LRGB technique, all images are taken in monochrome in FIT format. We will do the same during the acquisition of the images of the deep sky object, these images will also be taken in temporary black and white RAW mode or in monochrome with color filter for the LRGB technique. The subtraction or division of the preprocessing images from the images of the object will therefore be carried out in black and white, which will give a better result than doing it in color. After the subtraction, the resulting RAW (black and white) image is converted into color for color CCD cameras or APNs (digital cameras). For the LRGB technique, the LRGB images are stitched together to produce a color image. For full details of the LRGB technique, see section LRVB technique.

Here is the procedure to acquire these images :

1- The preload image (Offset or Bias)

The Bias represents the playback noise of the camera. This is a constant stray electronic signal that adds to the image of the deep sky object. The Bias is carried out by obstructing the focus of the camera and taking the shortest possible exposure time.

Here is an example of an offset image taken with the SBIG STL 11000 camera:

I increased the contrast of the image a lot to see the reading noise better. It can be seen that the signal is not uniform, it is darker towards the top of the image. We also see fringe effects (vertical light lines). Take 200 Bias images to later make a Bias Master using the compositing method Sigma clip.

2- The Black (Dark)

Black (Dark) is used to capture the electronic noise (thermal noise) of the CCD camera. It must be carried out with the same exposure time as the object to be imaged and at the same temperature. It should be noted that the Black contains the Bias (as all the images acquired, because it is the reading noise of the camera). So why take a preload image? If the black was not taken with the same exposure time as the deep sky image, the Bias must be used. Indeed, the Bias signal is always the same or constant. If using a black that was not taken with the same exposure time as the deep sky image, the auto preprocessing software will balance the black with the deep sky image, increasing or decreasing the black signal. . It will also increase or decrease the Bias signal which will result in not subtract correctly the Bias of the deep sky image (for more explanations, click on this link).

Black is taken by obstructing the focus of the CCD camera and by performing an identical exposure time for each composite image of the object to be imaged. Also, the black must be taken at the same temperature as the object to be imaged, so that the hot and cold pixels are identical. It can be taken before or after the astronomy session. Take at least 10 Black images to later make a Black Master using the compositing method Sigma clip. This will make it possible to eliminate the cosmic rays which can enter some black images as well as other parasites, especially when taking the black images immediately after the imaging session.

Regarding the temperature, it is recommended to use a camera that can regulate the temperature (for example, set a constant temperature at -20^o Celsius for deep sky images and blacks). This precision is necessary to ensure that hot and cold pixels are fully removed. Also, with each cooling of 6º Celsius of the camera matrix, the thermal noise decreases by 50%.

Personally, I realize my blacks in advance (during rainy days for example), so that I already have blacks with the same exposure time and at the same temperature as the object that I am. getting ready to image. I renew my banks of Blacks and Bias twice a year. The reason is that by using the camera regularly, the Bias and the Black change with the wear and tear of it. Regarding the PLU, it must always be taken after each astrophotography session.

Here is an example of a Black image taken with the same SBIG STL 11000 camera cooled to a constant temperature of - 10^o Celsius and with an exposure time ofe five minutess :

White dots represent hot pixels (Hot pixels ou Bad pixel) which always appear in the same place on the image for each composite photo. These hot pixels represent part of the thermal noise in the image. In this Black image, there are not many hot pixels due to the reason that the CCD camera is cooled to - 10^o Celsius.

In order to compare this Black image with the preload image shown in point 1 above, I increased the contrast of the Black image in the following example:

We can observe the vertical fringes of the Bias image. In addition, as in the precharge image, it can be seen that the signal is not constant. We also see a grainy structure in the more contrasted Black image. These are not additional hot pixels, but rather the dark current of the Black image that emerges in this very high contrast image. Hot pixels, vertical fringes, and dark current will disappear from the deep sky image when black is subtracted.

3- The uniform light range (PLU or flat field)

The uniform light range (PLU or flat field) corrects the vignetting problems of the optics used (the light tends to decrease in intensity as one moves away from the center of the image) and remove the dust that settles in various places on the optics and the camera sensor. It should be noted that we can try to remove all this dust, but there will always be some. Also the PLU makes it possible to correct the variations in sensitivity of the pixels of the camera.

The PLU is achieved by photographing a uniformly illuminated area such as a blue sky near the Zenith or any other uniformly illuminated surface. Take at least 10 PLU images with the same exposure time to later make a Master PLU using the compositing method Sigma clip.

Here is an example of a PLU made with the same SBIG STL 11000 camera mounted at the primary focus of the Takahashi FSQ 106ED telescope (iTeslescope-T14). The exposure time of each PLU image is one second:

I increased the contrast of the image to bring out all the flaws. We can see the vignetting very well (darker luminance at the edges of the image). We also see dust of different sizes appear. We also see the vertical fringes of the Bias, because the Black or the Bias has not been subtracted from this PLU image. On the other hand, we do not see hot pixels. The reason is that each PLU image, in this example, is one second and the camera is regulated (constant cooling) at -10^o Celsius. Hot pixels appear especially during long exposure times. Also the cooling of the camera allows to decrease the number of hot pixels.

The exposure time of the PLU

It is important not to saturate the image and also not to underexpose it. The rule is as follows for the following different types of cameras (CCD or APN):

To achieve the PLU, aim for a uniformly lit white background. Perform exposure time tests to arrive at the results in the table. The exposure time of the PLU should be at least one second to make sure not to record the movement of the camera shutter. Take at least 10 PLU images.

Here is the quickest method to check the saturation of the PLU. Immediately after having made an exposure time test, the histogram of the raw image which appears on the computer screen is displayed (for digital cameras, the histogram must be displayed on the camera screen ). Here is an example with the software Maxim DL. To display a portion of the histogram, select View | Screen Stretch Window. To achieve the PLU, aim for a uniformly lit white background. Perform exposure time tests to arrive at the results in the table.

Use the mouse to move the red triangle to the lowest value (Minimum) and the green triangle to the highest value (Maximum), as in the example shown. The image being in 16 bits, look in the table above, the saturation of the PLU must be between 22938 (35%) and 49152 (75%). Return to the histogram portion (Screen Stretch), the maximum value (green button) is 35835. It is between the recommended values. Then, look at the image of the PLU on the screen, we should see the dust and vignetting as in the following example:

We can see that the PLU is not overexposed. You can move the red histogram button to the left to decrease the image contrast and thus better appreciate the quality of the PLU:

By keeping the saturation (the maximum values) of the PLU between 35% and 75%, the preprocessing software will be able to calibrate the PLU according to the best values ​​for the raw image of the deep sky object to be preprocessed.

Here is another example with the Nebulosity software. The histogram shown is based on the same PLU image. To display the histogram, select View | Display control :

The histogram is displayed as a whole (0 to 65536 shades) instead of a small portion (Screen Stretch) offered by Maxim DL software. This makes it easier to validate the PLU at a glance. In fact, we immediately see that the maximum values ​​of the PLU are displayed near the center of the histogram. You can also analyze the quality of the PLU in the same way as Maxim DL. We uncheck the field Cars and we move the cursor with the mouse B (for Black - for darker PLU tones) to the left or right. You can also move the cursor W (for White - for the lightest tones in the PLU) as needed. At the same time, we look at the changes in the image displayed on the screen. Note that these operations will not affect the PLU images that will be saved on the hard disk of the computer.

For digital cameras (APN), the histogram is also displayed in full, like the Nebulosity software. To display the histogram of the PLU image to be taken, see the user manual of your APN. It should be noted that the majority of digital cameras can display a histogram of the image. With these devices, you cannot manipulate the histogram (there are no buttons or cursors to analyze the PLU). So, you just have to make sure that the maximum values ​​are between 35% and 75% of the histogram. Here is an example using my PLU box and my Nikon Coolpix S7000 camera:

The histogram is displayed at the bottom left. To adjust the PLU information to the center of the histogram, I increased the iris of the camera to +2 stops, as shown. Then, it remains to take at least 10 PLU images with this exposure time. Very fast and precise. By doing this, no need to do any other tests, it is certain that the PLU will be good.

Choose the diffusion surface

It is not easy to make a PLU because it must be produced with the same equipment that was used to take the photo and this equipment must not be reassembled or moved because the dust will no longer be in the same place for example. Focus should be the same as the captured image. So the ideal is to take the PLU immediately after taking the deep sky image. At this time, it is not easy to find a uniformly lit surface because we are in the middle of the night!

You can also take the PLU just before the imaging session. But, it may happen, during the imaging session, that we need to rotate the camera to properly frame the deep sky object in its field of view. Then, the PLU will no longer be good, because the dust will no longer be in the same place! Also, the focus may not be the same as the deep sky image.

Here is a simple method of producing a PLU image. It requires very little DIY. The PLU must be produced immediately after taking pictures to prevent dust from moving. To produce the PLU you need a uniformly illuminated surface. Due to the high sensitivity of the CCD camera, the light intensity may be low. The important thing is that it is uniform.

1- Use of a desk spot lamp equipped with a compact fluorescent bulb connected to a 110 volt outlet. If you are nomadic, it can be connected to an Eliminator battery which has a 110 volt outlet for example. We put a white cloth (sheet type) on the lamp (no danger, the fluorescent bulb is not hot).

2- Projection of the light beam onto a 3'x3 ′ white cardboard.

3- The telescope points at the white cardboard and its beam covers it completely.

4- The exposure time is determined manually to arrive at the correct saturation of the PLU (see section The exposure time of the PLU above).

5- Take at least 10 photos of the PLU.

The assembly is as follows:

The white cardboard should be parallel to the lamp for even light diffusion.

This method for carrying out a PLU is often used in a permanent observatory. The white cardboard is permanently installed (or any other matte white surfaces that withstand humidity well) on the wall or on the ceiling of the observatory. Then we also install the lamp permanently. You just have to do exposure time tests with your permanent equipment for each filter you use. As everything is permanently installed, during the next imaging sessions, it will be necessary to repeat these exposure times for each filter without having to do other tests.

The T-shirt method

Another simple method without DIY for to produce a PLU image in the middle of the night is to use a white T-shirt and an electronic flash. We put the T-shirt directly on the windshield mist of the telescope (put on 2 layers and make sure there are no folds). We start a pose of 10 seconds (the time to move to distance), stand about 5 meters (15 feet) from the telescope, point the flash in the direction of the telescope's field of view, and start the flash. We come back to examine the image on the computer to validate the light intensity.

The distance varies depending on the flash, equipment and filters used. So use this distance to start and then adjust it taking your equipment into account.

The box à PLU

You can also build a PLU (Flat Field) box inexpensively. Here are some interesting examples that I found on the internet:

• Rossell Croman's FLAT box (English)
• The boîte to FLAT by JF Brunelli

Personally, I built my PLU box based on JF Brunelli's plan.

Another (more expensive) option is to buy a light panel designed specifically to produce a PLU.

Note that the PLU box (as well as the light board) is the best method to produce a high quality PLU, as no ambient light will corrupt the PLU. For this reason, I advise you to use a PLU box to produce your PLUs.

Mathematical formulas for image preprocessing

Here is the formula that all astronomical image preprocessing software can put into practice:

• Calibrated image = (Raw image - (Black - Bias) - Bias) / (PLU - (Black - Bias) - Bias)

The calibration images are Dark Masters, PLU Masters and Bias Masters.

So, as the formula indicates, if we use the Bias (for the reason that the Black was not taken at the same exposure time as the deep sky image), it must be removed from the Black images. so as not to consider it twice, because the Bias is included in the Black.

In the formula, we use the Master Bias 4 times. Considering that each preprocessing image contains some level of noise which is in addition to the deep sky image, it is recommended to take 200 Bias images and make it a Bias Master. The remaining noise of this Master Bias will then be only 7,1% (see table on suggested exposure times), thus helping not to introduce too much noise into the deep sky image, so that the Master Bias is used 4 times in the formula, which will bring the remaining noise to 28,4% (4 x 7,1 %).

Many astrographers do not bother to make a black with the same exposure time and at the same temperature as the deep sky image. The reason is simple; when it comes to taking individual photos of 10 minutes or more, the realization of the black becomes a demanding task due to the time to spend to take the black as in the following example: 10 black images of 10 minutes duration each = 1,67 hours! If the camera does not have a regulated temperature, the task quickly becomes very demanding… So, in this case, we must use the above formula so that the Bias is correctly subtracted. It is also recommended to take a single set of blacks with its longest exposure time.

It should be noted that this preprocessing will not be as effective as taking Black with the same exposure time and at the same temperature as the deep sky image (there will remain hot pixels in the calibrated image and it will be more noisy because of the use of the Master Bias 4 times).

Considering that the Bias is alreadyà included in the Black, that the Black was taken with the same exposure time and at the same temperature as the deep sky image, we can simplify the formula as follows:

Calibrated Image = (Raw Image - Black1) / (PLU - Black2) Black1 = same exposure time and the same temperature as each of the raw images Black2 = same exposure time and same temperature as each of the PLU images

If the PLU images have an exposure time of less than 10 seconds (or 60 seconds for CCD cameras cooled to -10^o Celsius or more), we can ignore the subtraction of the image Black² of the PLU. On the other hand, it will be necessary to subtract the Bias from the PLU. In this case, the formula will be even easier to achieve and will be as follows:

Calibrated Image = (Raw Image - Black¹) / (PLU - Bias)

It is this last formula that I use for the preprocessing of my deep sky images. I use constant cooling at -20^o Celsius, summer and winter. It should be noted that each cooling of 6^o Celsius, the noise decreases by 50% and at this temperature (-20^o), there are almost no hot and cold pixels left on my PLU images. Also, when examining my PLU images, the noise is very low. It is therefore a very acceptable compromise compared to taking blacks with the same exposure time and the same temperature as the PLU images. This saves me the chore of producing specific black image banks for my PLUs. Also, as I use the Dithering during the acquisition of the images of the deep sky, it will be possible to make more easily disappear these few remaining imperfections in the PLU image during the compositing of the pre-processed images of the deep sky using the compositing method Sigma clip.

If possible, give preference to this pretreatment, as it will give the best results.

The number of images to take for each preprocessing image

To decrease the noise of preprocessing images, it is recommended to take multiple images. Here is an interesting formula to know (and already presented on the minimum suggested exposure times) to evaluate the remaining noise (compared to a single image):

% remaining noise = 1 / √ number of images

If we take 10 black images, the percentage of noise remaining will be 31,6%, ie a reduction in noise of 68,4%.

If we take 20 images, the percentage of noise remaining will be 22,4%, ie a reduction in noise of 77,6%.

So with 10 black images, the noise is reduced by 68,4% compared to a single image. If we compare this noise reduction with 20 images, the difference is 9,2% (77,6% - 68,4%). We must therefore judge the interest of the 9,2% gain. For example if the individual images of deep sky objects are 10 minutes long, if we take 10 blacks of 10 minutes, the total time will be 1h40. With 20 black images the total time is double that is 3h20. It's up to you to judge the benefit of the 9,2% gain in relation to your exposure time per photo. Another example; if the exposure time per photo is 2 minutes, a larger number of Black images can be made (the total time will be only 40 minutes for 20 Black images). It is also necessary to consider that an image taken with an exposure time of 2 minutes is more noisy than that taken with an exposure time of 10 minutes because the difference between the signal and the noise (S / N) will be less. than that of 10 minutes.

For PLU images, it is easy to take more than 10 images because the exposure time per photo is short (most of the time less than a minute for each PLU image). For example, if each PLU image is 10 seconds, taking 50 PLU images will be only 8,33 minutes long and noise will be reduced by 85%. For Bias, it is recommended to take 200 (see explanation above).

Creation of Bias, Black and PLU master images with Maxim DL

Here is how I go about creating the Bias, Black and PLU master images using the Maxim DL software. If you are not using this software, you can use this procedure as a guide to create your own master images using your favorite preprocessing software.

Create a Master Bias image:

• Process | Stack ...
• Deselect Auto Calibrate, Auto Color Convertert and Make Pixel square
• Click on the button Add Files ... and select all the images to be composed.
• In the tab Quality, deselect all Criteria
• Click on the button Aling
• Mode: None
• Click on the button Combine
• Combine Method : Sigma-Clip
• SigmaFactor : 3 or 2 (do both)
• Select ignore Black Pixels
• Normalization : None
• Click on GO
• Click on the button Close
• Queue | Save : file name for example Master Bias - Bin 1 × 1 -20 Celsius-Sigma 3
• Compare the Sigma image 3 and 2 and choose the best image. Always favor the Sigma 3 image (which rejects fewer pixels). If there are cosmic rays or other parasites present, then choose the Sigma 2 image.

For Master Bias, the images should not be normalized. Until August 2019, I used the Average compositing method to produce my Master Bias. But, in August 2019, during the up-to-date production of my Bias Master, there was a cosmic ray that entered one of my Bias images. Using the Sigma Clip compositing method with filter 3 (Sigma Factor above), it was removed. For this reason, I now recommend using the Sigma-Clip compositing method to produce the Bias Master.

Create a Black Master image:

• Process | Stack ...
• Deselect Auto Calibrate, Auto Color Convertert and Make Pixel square
• Click on the button Add Files ... and select all the images to be composed.
• In the tab Quality, deselect all Criteria
• Click on the button Aling
• Mode: None
• Click on the button Combine
• Combine Method : Sigma-Clip
• SigmaFactor : 3 or 2 (do both)
• Select ignore Black Pixels
• Normalization : None
• Click on GO
• Click on the button Close
• Queue | Save : file name for example Black Master - Bin 1 × 1 - 600 sec -20 Celsius-Sigma 3
• Compare the Sigma image 3 and 2 and choose the best image. Always favor the Sigma 3 image (which rejects fewer pixels). If there are cosmic rays or other parasites present, then choose the Sigma 2 image.

For the Black Master, one should not normalize the images.

Create a PLU master image:

• Process | Stack ...
• Deselect Auto Calibrate, Auto Color Convertert and Make Pixel square
• Click on the button Add Files ... and select all the images to be composed.
• In the tab Quality, deselect all Criteria
• Click on the button Aling
• Mode: None
• Click on the button Combine
• Combine Method : Sigma-Clip
• SigmaFactor : 3 or 2 (do both)
• Select ignore Black Pixels
• Normalization : Linear
• Click on GO
• Click on the button Close
• Queue | Save : file name for example Master PLU - Bin 1 × 1 - Blue - 20 sec -20 Celsius-Sigma 3
• Compare the Sigma image 3 and 2 and choose the best image. Always favor the Sigma 3 image (which rejects fewer pixels). If there are cosmic rays or other parasites present, then choose the Sigma 2 image.

For the Master PLU, as for the images of the deep sky, it is necessary to normalize the images. Normalization is used to set the same level of intensity to all pixels in the image.

Richard Beauregard
Sky Astro - CCD

Revised 2022/08/21

References :

For recommendations on whether or not to use Bias in the preprocessing of deep sky images:

The New CCD Astronomy, Ron Wodaski, New Astronomy Press, pages 244 and 245.

Usefulness of the preload image (Bias or Offset), detailed explanations in this site.

For recommendations and detailed explanations of using the Sigma-Clip compositing method for Black and PLU:

Preprocessing or calibration of CCD images, Denis Bergeron.

Comparative analyzes of the different compositing methods. Advantages and disadvantages presented in this site:

The different compositing methods

Mathematical comparison between mean, median and Sigma-clip