CCDIS Plug-in for CCDStack

A new and exciting development in CCDInspector 2 is the inclusion of an advanced, automated image registration module that works seamlessly with CCDStack. This module, CCDIS will only work if both, CCDStack and CCDInspector are installed on the same computer.

Image registration is the process of adjusting an image stack so that all the images are properly aligned to a base image. This can involve scaling the images to the same dimensions, shifting them vertically and horizontally, and de-rotating them so as to allow them to be stacked.

Without CCDIS registration process can be tedious, and the resulting registration may not be as accurate. Inaccuracies in registration can lead to blurring and loss of resolution in the stacked image -- something to be avoided at all costs! CCDIS uses an extremely accurate and sophisticated algorithm to find unique star patterns between all the images, automatically! CCDIS then applies the appropriate scaling factor, de-rotation, and linear offset to align the images in the best possible way. CCDIS will also flip the image if the horizontal or vertical axis are reversed. Generally, CCDIS alignment accuracy will be better than 1/10 of a pixel. In many cases it is better than 1/30 of a pixel!

CCDIS plug-in is included and automatically enabled during the installation of CCDInspector 2. If you already have CCDStack or decide to install it later, CCDIS will quickly become your primary choice of registration methods:

When installed, CCDIS tab will be the first tab of available registration methods in the CCDStack Registration dialog. There are no complicated settings: just click align all or align this buttons to perform the fully automatic registration. When done, the results of the registration will be shown in the lower half of the dialog.  Results for this image section shows the registration statistics for the currently selected image. Above example indicates that a total of 325 stars were chosen for the registration process, and from these, a 23-star unique pattern was identified that matched with the base image (base image is shown on the bottom of the dialog). The RMS error shown is the average error across all 325 stars, after the registration process completes. This is expressed in pixels. The above result is very good, showing that the error is almost 1/50 of a pixel!

The only option in the CCDIS Registration dialog is to use high precision. This option makes CCDIS do more work by extracting more stars from the image, and by looking for the star pattern for a bit longer. The results are usually somewhat improved compared to the low-precision registration. High precision mode can take a bit more time, but can work better especially for images that do not have a large overlap, such as when doing mosaics. If CCDIS is unable to find a matching star pattern in the low-precision mode, it will automatically switch to the high-precision mode and try again. In most cases, it is safe to leave use high precision option unchecked.

Once registration completes, you can see the actual transformation settings CCDIS found for each of the images by opening the Manual tab:

As you select different images in the image manager, the values in the Manual tab will change to show the CCDIS-determined registration parameters. In the example above, the selected image was shifted over by -69.13 pixels horizontally, 237.52 pixels vertically, and was rotated by 33 degrees. The image scale was found to be the same, so the scaling factor (Magnify) was set to 1.000.

NOTE: after performing registration and looking at the color display in CCDStack, the stars may appear to be slightly out of alignment, especially at high magnification. This is because the display cannot adjust images by a fraction of a pixel, and is not an indication of CCDIS error. Once you apply the registration to the images, the position errors will disappear (except when using the Nearest Neighbor interpolation method, the stars also cannot be adjusted by less than a whole pixel).

CCDIS uses a highly optimized, efficient algorithm to perform pattern matching. It will use multi-core, or multi-processor computers to perform its functions even faster than on a single CPU.