General linear transformations

To align datasets that require more general linear transformations than a simple offset use the:

• CCDALIGN

procedure. This accepts either a sequence of images or related images (datasets which are already approximately aligned, at least within the capabilities of centroiding $\approx$ few pixels) and displays them one by one (or the first member of each group). You then have to simply identify the image features to use, but in the correct order. Only enough image features to identify the approximate image position are required as the procedure then centroids the image features, works out an approximate registration which it then uses to extrapolate the positions of a reference set on each dataset. This new extended set of positions are now centroided picking up any missed objects. The reference set of positions are either selected from a designated image or from the first image. CCDALIGN can then invoke REGISTER itself to perform the fitting of the transformation parameters using all these positions.

Selection of points on the images is done using an inuitive Xwindows-based graphical user interface, which allows you to mark and remove points with the mouse, change the brightness of the display, zoom and scroll the image, and read out positions in any of the image's attached coordinate systems.

To summarise, during the graphical part of CCDALIGN you have to do the following:

1.

Mark a set of points on the reference (or first) image

2.

Mark corresponding points on each of the other images