Installation

Menu Structure

Taking Better images

Other programs

Colour Programs

Introduction

The software was written for a rather under powered computer. The requirements were to produce a program the would run as fast as possible on a DOS computer with a 33mhz 386 PC with 4Mb of RAM. For speed it was written in the C programming language. Single keystrokes are used for most functions. The windows interface was not used - this would have made the program intolerably slow. The routine to capture images from the frame store has been highly optimised. The program also automates many boring functions such as continuous downloading and flat field accumulation. A separate program has been provided for the processing and alignment of a number of images, dark fields and flat frames. The program should work under Windows 3.11 and Windows 95.

Software & Installation

release: the software consists of two files, ccd.exe and ccd.ini. Create a directory C:\PICFILE and place all files therein. Edit the file CCD.INI using a text editor.

All the required files as well as a copy of this manual can be downloaded here as a .ZIP file

smooth=1
tp_directory=C:\PICFILE
fits_directory=C:\PICFILE
ps_directory=C:\PICFILE
tiff_directory=C:\PICTIFF
bmp_directory=C:\PICFILE
sb4_directory=C:\PICFILE
bmp_xpixels=1024
bmp_ypixels=768
disk_save_type=5
disk_load_type=5

running the program. run the program ccd.EXE from DOS or Windows. You may supply an optional filename, if the program can open this file as an image file then the program will go straight into display mode with this image.

note: the ccd.ini file must be placed either in the current directory or in the directory 'C:\WINDOWS'. If the file cannot be found then default values will be assumed.

Documentation

Menu structure

Disk Menu -------------------------------------- :D

Flat Field Menu -------------------------------- :F

Contrast Stretch ------------------------------- :C

Miscellaneous Menu ----------------------------- :M

Histogram -------------------------------------- :H

Show Image ------------------------------------- :S

Exit program ----------------------------------- :X

Disk Menu

Flat Field Menu

Contrast Stretch Menu

Miscellaneous Menu

Histogram

The title on the top of the graph shows the filename (if applicable), the max, min and average value of all pixels in the area being measured. This area is not the entire image but the square from (128, 64) at top left to (384,192) at bottom right - this is the middle half of the image.

The screen capture above shows the histogram of an unprocessed image of the Virgo galaxy M87. The X axis is labelled with brightness values from 0 to 4095. The peak value on the graph at around 400 corresponds to the average sky brightness - the most common brightness value on the image. The section of the graph to the right of this peak shows the object part of the image.

Show Image

Load File -------------------------------------- :L

Save File -------------------------------------- :S

Erase File ------------------------------------- :E

Fetch from Framestore -------------------------- :F

Push to Framestore ----------------------------- :P

Change Directory ------------------------------- :D

Change Load File Type -------------------------- :C

Change Save File Type -------------------------- :T

Exit Menu -------------------------------------- :X

Load File

Save File

Erase File

Fetch from Framestore

Note: if this option is run at the same time that another image arrives from the CCD head into the Framestore then the two operations interfere with each other resulting in a corrupted image. 

Push image to Framestore

Change Directory

Change Load File Type

• 0 - TP or Terry Platt format, this was the original file format where every image was split into four files '#file', 'file2', 'file3' and 'file4'. This image type should now be redundant. Note: an automatic file converter from this format to the new format has been written by the author.
• 2 - fits. the flexible image transport system. An industry standard file format designed for scientific use. The header information of the file may contain information including: date, time, number of exposures, dark frames, flat fielding, filters, exposure, gain level, observers and comments in fact anything that you want. This may be the preferred file format in the future. Save files in this format if you wish to use other scientific image processors for further processing.
• 3 - frame, to load from the frame store. This option is really redundant because you can use the fetch from frame store option above.
• 5 - new TP - the new format files as used by pixwin store a whole image in just one file. Images are stored in 16-bit form.
• 6 - SBIG - to read images from the SBIG ST-4 camera.

Change Save File Type

• 1 - postscript format for printing to a postscript laser printer.
• 4 - bmp or bitmap. Save the image as a 256-colour bitmap file measuring 1024 by 768 pixels. Note: you can change the size of the saved file by editing the program's configuration file ccd.ini. The image is saved in the higher resolution format by resampling. Note this option can be called from the 'Show Image' facility where you also have the option of saving zoomed parts of the image.

Flat Field Menu

Subtract Dark Frame ---------------------------- :S

Flat Field Image ------------------------------- :F

Add Images ------------------------------------- :A

Mask Image ------------------------------------- :M

Integrate Dark Frame --------------------------- :D

Integrate Image Frame -------------------------- :I

Continuous Capture ----------------------------- :C

Exit Menu -------------------------------------- :X

Subtract Dark Frame

this option is normally used in the following way. Take your object image and save onto disk, take or locate a dark frame at a similar temperature and identical gain and exposure settings. For more hints on taking better images follow this link. 

Flat Field Image

Before calling the routine load the image file and subtract the dark frame. The routine will prompt for the name of a flat file. The flat file is divided into the image to flatten the field.

How is it done? A good flat field file will have pixel values somewhere in the range 2048 to 4095, one of my fields has a max of 2770 a min of 2553 and an average of 2673 brightness. The noise of the image is measured by the Statistics program at 0.337 percent - fairly smooth, the range in pixel brightnesses is 223 over the average brightness of 2673, thus pixels in the flat field can differ by as must as ten percent so this correction is certainly worthwhile. To speed up processing the calculations are performed using integer arithmetic, to prevent clipping the calculations we use integers that can store from 0 to 4 billion. The image file pixel value is multiplied by 4096 and then divided by the flat field pixel value. In our example file an image pixel value of 3000 would be multiplied by 4096 and then divided by say 2673 giving a value of 4597. The effect of flat fielding is to brighten the image as well as giving a uniform sky background. If you wish to reduce the brightness back below 4096 then you can use the Multiply image facility. 

Add Images

Mask Image

Next load the long exposure image from the disk menu, use this mask option and enter the mask filename. Use 'Show Image' to look at the resulting image, the brighter parts of the long exposure should have been set to black. Finally use the 'Add Images' option mentioned above to add in the short exposure and look at the results. 

Integrate Dark Frame

Run the option, enter a filename such as d22a123 (for perhaps dark, 22 degrees below, a for averaged, 12 for exposure of 12 (80 seconds) and 3 for a gain setting of 3). Do not enter more than seven characters. Then enter the number of dark frames that you wish to capture. Then hit return when the display is showing a good dark frame. The program will then continually read the images as they appear on the frame store and add them together into the current image stored in the computer. The program remembers the statistics for the first image. With each frame capture the statistics of the image are measured and displayed, if the average pixel brightness differs from the first image by more than 6 percent then the message 'rejected' is displayed and the computer waits for the next image. Otherwise the image is accepted. The statistics displayed for subsequent images are as follows: the frame number, the average brightness and noise of the last image, the maximum value and noise in percent of the accumulated image. As time progresses the noise, as a percent of the image should fall to around 0.3 percent, beyond a certain number of images little improvement in the noise will be noted. When the correct number of images has been captured or the brightness of the brightest pixel has reached 60000 the program will exit the loop.

You are now prompted with the question: 'Do you wish to store average or sum of the frames? (A/S):'. If you enter 'A' the computer will divide the brightness of each pixel by the number of frames successfully captured. This stores a smooth dark frame that can be directly subtracted from an image taken at the same settings. You may sometimes wish to store the sum of the frames by entering 'S' - you will do this when trying for the best quality images where you do not wish to have any rounding errors. 

Integrate Image Frame

The program is not good at detecting bad frames (clouds, car headlights or tracking errors). And so is not recommended. If you wish to try it then follow the instructions on the screen.

There is a better option to this program: use the continuous capture program to store image files in consecutive files, delete any images that are not suitable for integration. Use the same option to store flat fields. Use the integrate dark frame routine to store an averaged dark frame. Process the flat fields with the dark frame to get a smooth flat field. Finally use the separate 'offset' program to generate the final image file offsetting images where required. 

Continuous Capture

The program will not work for exposures of 4 seconds or less - there is not enough time to detect the new image and download it before the next image comes down from the CCD.

How does the program now that a new image has been loaded from the CCD into the Framestore? Once an image has been downloaded a pixel on the bottom of the image is marked (set to another value). Periodically this pixel is checked to see if it has been overwritten. Once it has been overwritten the program assumes that a new image has arrived and it begins to copy it into the computer. 

Contrast Stretch Menu

Normal Stretch --------------------------------- :N

Auto Stretch ----------------------------------- :A

Log Stretch ------------------------------------ :L

Power Stretch ---------------------------------- :P

Unsharp Mask ----------------------------------- :U

Row Display ------------------------------------ :R

Column Display --------------------------------- :C

Exit Menu -------------------------------------- :X

Normal Stretch

Autoscale

The computer cannot use the entire CCD image to calculate the highest and lowest pixel values because there may be borders on the image. Instead it uses the area from (128,64) to (384,192) which is approximately the center half of the image.

Note: this routine does not work too well when the image has a faint object but a bright star. Generally the star saturates at 4095 and the routine just stretches the sky background down to black.

A histogram is shown at the end of the stretch. 

Log Stretch

Power Law Stretch

Unsharp mask

Row Display

This image is an unprocessed image of M87 as used for the histogram and show image descriptions. The Y axis is measured in thousands. The sky background can be measured at around 400 and the peak brightness is around 2100.

Whilst the profile is displayed a number of keys are active to change your view. N and P take you to the next or previous row. Z zooms into the central half of the screen. Left and right arrow keys move the displayed area of the line move (only if already Zoomed). Any other key will drop you back to the stretch menu. 

Column Display

Miscellaneous Menu

Exit Menu -------------------------------------- :X

Auto Focus

Multiply Image

Image Statistics

Max - the brightness level of the brightest pixel in the area.

Min - the brightness level of the darkest pixel in the area.

Avg - the average brightness of all the pixels in the area.

Sdx - a measure of the average difference brightness between adjacent pixels.

Sdx2 - a measure of the average of the square of the differences between adjacent pixels - the image sharpness.

Noise - the ratio of the difference between adjacent pixels to the average value of the pixels. If the average difference between adjacent pixels is 4 brightness units and the average brightness of a pixel is 600 then the background noise is 4/600 = 0.67 percent - a good noiseless image. As you take more images for stacking this noise level will decrease. If you are measuring this value after dark framing and flat fielding you will be measuring the true quality of the image and the lower the value the more the image can be stretched to show fainter details. Hit any key to return to the menu. 

Clip Image

Invert Image

Clear TP Header

Edit Image Contents

Show Image

this image is a raw image of m87. The sky brightness can be seen as the grey sky background. This is the same image as used for the histogram and row display explanations. The core of the galaxy is visible along with three other galaxies and the active 'jet'.

Once the image has been displayed a number of keys become active to perform various functions. Press the X key to leave the show image function and return to the main menu. To see a list of these keys press the ? key. These functions and keys are described below:

• z - zoom in to the central half of the screen.
• z - zoom out again.
• i - inverse image. White becomes black and vice versa.
• s - toggle smooth display.
• Cursor keys - only when looking at a zoomed image. Move a quarter of the screen width in the direction of the cursor key.
• INSERT KEY: - toggle scroll distance from a quarter of a screen to individual pixels.
• c - perform an automatic contrast stretch. Note: this does not always do what you expect, if you have a galaxy next to a saturated star image then the program will generally remove the sky background but will not stretch the image at the already saturated end and so the image will appear darker overall. In these cases you will need to go to the contrast stretch menu to perform a manual stretch.
• P - push this current image up to the frame store. The frame store screen will frequently show images better than the computer monitor. This function even works when the Smooth switch has been selected and also the zoom facility.
• F - set colour palette to false colour.
• b - output to bitmap file. This option creates a file called screen.bmp in the current directory. The size of the bitmap will be controlled in the CCD.INI file. By default the bitmap file will be 1024 x 768. The bitmap will be calculated by resampling the image to the larger size.
• M - photometry routine - see below

Photometry Routine

• Cursor Keys - move the cursor on the screen by the current scroll distance.
• INSERTKEY - toggle scrolling speed from 1 to 10 pixels.
• PAGEUP - increase area of the cursor up to a maximum of 40 by 40.
• pagedown - decrease area of measurement, i.e. a smaller cursor.
• Enter - Display the sum of the brightness values of all the pixels under the cursor.
• '?' displays the x and y position of the point under the cursor.
• 'X' return to the screen display menu.

Taking Better images

• Fix the CCD head onto the telescope, plug in all of the cables.
• power up the computer and the ccd. Set the fan and cooler onto high cooling.
• Whilst the CCD is cooling, open up the dome and telescope fully place the focussing aid over the end of the telescope. This is an aperture stop for the telescope with two holes cut in the aperture to aid focussing. Set the camera on a 4 second exposure any faster and the autofocus programs do not work. Place a star just to the upper left of the center of the field. Select Auto Focus 4, the star should be seen greatly enlarged on the computer screen. Move the CCD through the focus and back to get the best focus.
• locate your first object. Check the temperature to see if it has become stable. Find out an exposure that gives good image on the screen. An exposure of 12 (or 80 seconds) with a gain of 3 is usually ideal. If the tracking is good or the object has a high declination you might use exposure 13 (160 seconds). The image should have a well defined background level from between 100 to 1000, the brightest objects should not be clipped (unless you are not interested in them).
• fit a dark mask over the ccd. Flush out the current image and wait for the first dark frame. Go to the Flat field menu and select 'Integrate Dark Frame'. We are going to take 8 images of the object for stacking so we need to take 8 dark frames. If the temperature of the CCD has not yet settled down you may wish to take 4 dark frames now and 4 at the end of the observing run and average them. If the CCD temperature has stabilised then enter d22a123 for the filename (in this case 22 is for -22 degrees, 12 and 3 are for the exposure and the gain settings, d means a dark frame and a means averaged. We can use this dark frame on other nights where the conditions are the same. Enter 8 for the number of frames that you wish to capture.
• Wait for the images to be captured, in this case this will take 8 * 80 = 640 seconds or just over 10 minutes - during this time have a cup of tea or plan the observing session. You will see the frames being downloaded as they arrive in the framestore. Note that the noise of the first image will be of the order of 1.2 percent, this is fairly independent of exposure or gain. The noise will decrease as more images are added together, it will converge to a noise level of around 0.25 percent, it will never reach zero. The column labelled 'Average' shows the average brightness of the CCD frame, if this is changing steadily it is a sign that the CCD temperature is changing, if the value is decreasing then the CCD is cooling and vice versa. Once the Average value has changed by more than 6 percent from the first frame the images are rejected. Once you have captured the eight images you are prompted to store the A/S average or sum of the images in the file. Choose A, storing the Sum would potentially be slightly better but the difference is not too important at this level.
• Now let us use the Continuous Capture facility to capture our images. Remove the mask over the CCD head, flush the current image and wait for the first image to arrive. Enter say 'GH' as the filename prefix and hit return. Watch as the images are captured and stored as '~GH1', '~GH2'. '~GH3', '~GH4' etc … You can either have another cup of tea or watch the individual images to see if some need to be rejected. After 8 or 10 images have been stored, examine each image for quality - cloud, bad tracking etc. Store more images if necessary.
• If the sky might cloud over then take the images at the start of the night and take dark frames and flat fields later, these can be done when it is overcast.
• This step can be skipped if time is more important than quality. It is the storing of the Flat field. Again choose continuous capture. This step is described in a section below, the result is an image on the disk called '~flf4', this naming for a flat field for the F4 focus of our telescope, other flat field names could be fl200mm for a 200 mm telephoto.
• now exit the ccd program. Run the command: 'offset gh d22a123 flf4 gh1 gh2 gh3 gh4 gh5 gh7 gh8 gh10'. You notice that gh6 and gh9 are missing, this is because they were deemed of too poor quality. The computer reads all of the 'gh' files and creates '~GH'. Examine this image using the CCD program you should have a sky background that is uniform and with a very low noise that can take a great deal of contrast stretching before the noise becomes apparent. If you skipped the flat fielding phase and entered 'null' instead of 'flf4' in the offset.exe program then you may see dust effects and stripes going vertically across the image. Only flat fielding can remove this effect.
• If you have taken a wide field exposure with a camera lens you may have noticed that the image is brighter on one side than the other. This is usually due to variations in sky brightness across the field. Curing this problem can only be done with a more sophisticated program such as MIPS to create a grid of theoretical sky brightness and then subtract this surface from the whole image.

Creating a Flat Field Image

ensure that the ccd's temperature has stabilised. You need to image an evenly illuminated sheet at full aperture. The sheet should be illuminated by white light rather than monochrome light. For small apertures you can fix a tracing paper like mask over the end of the telescope and illuminate it evenly. Setup the CCD to take short exposures (around 10 seconds each) at a gain setting such that the average brightness of the pixels is between 2000 and 3500, this is important to reduce noise in the finished flat field. Use the Continuous capture facility to record perhaps 16 of these fields, lets say called df1, df2 … df16. Now remove the mask and take 16 dark frames using the Integrate Dark field and store the sum of the frames the 'S' option rather than the average of the frames. The background count of these dark frames should be far lower than the 2000 - 3500 of the 'df' frames, store the Integrated dark frame in the filename 'dtemp'.

Now load 'gh1' into memory, from the Flat Field menu use the 'Add image' option to add 'gh2' then 'gh3' and finally 'gh4' and store in filename 'gh1-4'. Repeat this process adding 'gh5' .. 'gh8' together into file 'gh5-8' and likewise create files 'gh9-12' and 'gh13-16' - doing it this way will reduce errors. Finally load 'gh1-4' into memory and add to it images 'gh5-8', 'gh9-12', 'gh13-16' and store the total into 'ghtotal'. Now use the 'Subtract Dark Frame' option to subtract the file 'dtemp'. We now have a flat field file. Run the 'Image Statistics' option on the file to see the average brightness of the file, lets say the average brightness is 34724 (remember we have added up 16 images together). We should really reduce this smooth flat field back down to a 12 bit image and so we now use the 'Miscellaneous menu' option of Multiply image and enter a value of 0.1, this gives us a flat field with average values of around 3470. Save this flat field now as 'flf4' in our example for flat field at F4.

The last and important step is to delete all of the temporary image files 'gh1' .. 'gh16', 'gh1-4' etc. This good housekeeping is essential.

Use the Image Statistics to examine the noise of the flat field - it should be around 0.3 or 0.4 percent. Look at the image with 'Show Image' it should seem very even, perform an Auto Contrast stretch on the image by hitting 'C'. Your should now see in great detail the patches on your CCD setup as well as the vertical lines of differing sensitivities. 

Other programs

usage: offset.exe output_file dark_frame flat file1 file2...

All files are assumed to be in the new 'pixwin' format i.e. begin with a '~'. The output file is the file to be created as the sum of all the image files. The dark_frame is a dark frame taken at the same settings as the image file, this should be an averaged dark frame, if you do not use an averaged dark frame the output file will be very noisy. If you do not have a dark field you may substitute the word 'null' for this parameter. The flat file is a flat field image, if you do not have a flat field image you may substitute the word 'null' for this parameter. All of the remaining parameters are files to be added together.

At present the program assumes that all files are to be found in the directory C:\PICFILE . This is a bit of a nasty assumption which should be changed in a future version.

offset2.exe: similar to offset.exe but with a different algorithm for aligning the images.

convall.exe: This program converts all Old Terry Platt format files: #file in the current directory into the new ~file format files.

You must ensure that you have sufficient disk space before running the program, you will need free space equal to the amount of files in the directory.

Note: the date and time of last file modification are unchanged by the program. You are prompted to hit 'Y' before the conversion proceeds. The program does not remove the old format files from the directory, to do so would be extremely presumptious, you will feel more confident doing this yourself.

push.exe: Used for displaying a number of images on both the computer screen and the frame store monitor. The first parameter is the interval in seconds between displaying files. The files to be displayed must all be in the new 'pixwin' format, i.e. beginning with a '~'. Once the last image has been displayed the program goes back to displaying the first image once again. The program can be aborted by hitting the 'Q' or 'X' keys. All image files must be in the directory c:\picfile.

usage: push.exe interval m1 m15 m74 m31 ng7331c etc… 

Colour Programs

Ensuring a correct colour balance is far more difficult than it might be thought. It is important to ensure that the flat fields were taken using white light and not sky or clouds illuminated by orange sodium lights.

The sky background is normally subtracted from images. The sky background will be dominated by the orange glow from local lights, the background will therefore appear differently on each filtered image. After the images have been processed for dark frames and flat fields, measure the background sky brightness on each image. Lets say the sky background has a value of 200 on the red image and only 130 on the green image. We need to stretch the red image using the normal stretch using a low value of 200 and a high value of 4095+200=4295. The green image is similarly stretched from 130 to 4095+130=4225, this means that we have allowed for differing sky backgrounds without increasing the brightness of one colour against the next.

The next procedure will be to use the 'Multiply Image' to alter the brightnesses of the various images before collating all three images.