I have a scope that, like many, produces very good images on axis, but which has obvious elongation (coma) far off axis. This is particularly apparent when used with a
camera chip.
In the images shown here the stars had (in the center of the field) a FWHM of about 3,5 pixels. In the extreme corners the stars were elongated to about 4x9 pixels. I.e. there
was a distortion of about 5 pixels in the long axis.
To correct this I first made a copy, then resized that copy slightly. Since the stars were elongated by about 5 pixels in the corners I decreased the size of the image along the
diagonal by 10 pixels, shrinking the radial from the center to each corner by 5.
I now had the original image, and the resized image. In both the off axis stars were ovals. In the corners of the image the ovals were offset by about 5 pixels and were about
9 pixels long so there was a 3 to 4 pixel overlap. Progressively towards the center of the image the elongation was less, but the effect of the resizing was less too. Half way
from the center to the corners the ovals were less elongated but only offset (relative to each other) by about 2.5 pixels. Near the center where the stars were round the
resizing had almost no effect.
See animated GIF panels “Original” (the original image) and “Contracted” (the resized image). Note this is a crop of the full frame, taken from the lower right corner.
camera chip.
In the images shown here the stars had (in the center of the field) a FWHM of about 3,5 pixels. In the extreme corners the stars were elongated to about 4x9 pixels. I.e. there
was a distortion of about 5 pixels in the long axis.
To correct this I first made a copy, then resized that copy slightly. Since the stars were elongated by about 5 pixels in the corners I decreased the size of the image along the
diagonal by 10 pixels, shrinking the radial from the center to each corner by 5.
I now had the original image, and the resized image. In both the off axis stars were ovals. In the corners of the image the ovals were offset by about 5 pixels and were about
9 pixels long so there was a 3 to 4 pixel overlap. Progressively towards the center of the image the elongation was less, but the effect of the resizing was less too. Half way
from the center to the corners the ovals were less elongated but only offset (relative to each other) by about 2.5 pixels. Near the center where the stars were round the
resizing had almost no effect.
See animated GIF panels “Original” (the original image) and “Contracted” (the resized image). Note this is a crop of the full frame, taken from the lower right corner.
| Elongation Eradication |
| Top |
 |
 |
Now in CCDStack I did a Minimum Combine of the "Original" and "Contracted" images. Minimum Combine uses the least value at each pixel location for the final image. Where
the ovals did not overlap the value for empty space was used. Where the ovals overlapped a high value consistent with the star was used. This resulted in bright (star) values
only where the ovals overlapped.
See “Min Combine” in the animated GIF.
The stars now look pretty round. If you look in the "Original" image the ovals towards the top of the frame are elongated right/left. Towards the bottom of the frame the
elongation is more diagonal because the elongation points towards the center of the field. But the contraction due to resizing is also towards the center of the field along that
same axis, so the correction both at the top and the bottom of the image produces round stars.
Note there is one bright area (marked by the arrow) which is not elongated along the same axis as the other bright areas. Because this object does not have a radial elongation,
oriented to the center of the field, the offset and Min Combine does not make this object round but makes it a more vertically oriented structure. That object isn’t a star. It’s a
galaxy - NGP0F378-0022175. See the Palomar Observatory Sky Survey image. That galaxy does, indeed have a somewhat vertical orientation as shown in the POSS image.
So this correction method salvages objects other than round stars. It should. The distortion is symmetric and regular so objects of any shape are returned to something closer to
their non-distorted appearance.
the ovals did not overlap the value for empty space was used. Where the ovals overlapped a high value consistent with the star was used. This resulted in bright (star) values
only where the ovals overlapped.
See “Min Combine” in the animated GIF.
The stars now look pretty round. If you look in the "Original" image the ovals towards the top of the frame are elongated right/left. Towards the bottom of the frame the
elongation is more diagonal because the elongation points towards the center of the field. But the contraction due to resizing is also towards the center of the field along that
same axis, so the correction both at the top and the bottom of the image produces round stars.
Note there is one bright area (marked by the arrow) which is not elongated along the same axis as the other bright areas. Because this object does not have a radial elongation,
oriented to the center of the field, the offset and Min Combine does not make this object round but makes it a more vertically oriented structure. That object isn’t a star. It’s a
galaxy - NGP0F378-0022175. See the Palomar Observatory Sky Survey image. That galaxy does, indeed have a somewhat vertical orientation as shown in the POSS image.
So this correction method salvages objects other than round stars. It should. The distortion is symmetric and regular so objects of any shape are returned to something closer to
their non-distorted appearance.
allow precision astrometry. It's an approximation. But it's a big help in repairing images distorted by off axis coma.
To see the full image used in these examples click on the thumbnail below. Images taken with a Vixen VMC260L OTA and SBIG STL11000M camera.
To see the full image used in these examples click on the thumbnail below. Images taken with a Vixen VMC260L OTA and SBIG STL11000M camera.