The DataFlow environment on the VLT ensures that data such as sky position pointed to by the telescope should be present in every FITS header. However, no telescope pointing can be precise up to subpixel precision and expected errors are in the range of a few pixels. For this reason, telescope offsets are taken as a first estimate of the place to look for, but not as an absolute reference. This estimate is fed into the offset precise detection algorithms described hereafter. Users can also provide their own offset estimations if they have other tools to compute them. Last, if no first estimate is provided, a default null vector will be used, with a larger correlation search domain.
Cross-correlation is a very efficient tool to match images. It is quite robust to noise, and can be normalized to allow pattern matching independently of scale and offset in the images. It has serious drawbacks however, in the case images of faint sources. We discuss here the implementation of a cross-correlation tool, the parameters which are used to configure it, and try to define which are the best suited parameters for a given search.