Registration and mosaicing

Multiple observations form the backbone of many astronomical programmes. Determining the registration (inter-dataset transformations) of the observations is a necessary step when preparing to inter-compare or combine the data. Inter-comparison is used when performing multiple waveband observations; combination when measurements beyond the capabilities of the detector are required. Helping astronomers to determine the registration of imaging data and subsequently to transform positions or resample and combine the data is the purpose of this part of CCDPACK.

A number of registration techniques are provided, which can identify the relative positioning of frames by examining image features (centroidable objects), or by making use of prior information about the observational geometry, or both.

If it is the intention to combine datasets into one (to increase signal to noise levels, or to increase the effective area or dynamic range of the detector) the registering transforms may be used to resample the datasets so they are aligned (i.e. have pixel-to-pixel correspondence). If atmospheric transparency, sky brightness or exposure times have varied between the datasets, they need to go though a process of `normalisation', in which global zero points and scale factors are determined. After these stages the data may be combined to produce a mosaic. Data combination usually makes use of a robust estimator to protect against spurious values, cosmic rays etc.

Version 3 of CCDPACK, which was released in mid-2000, handles the registration process differently from previous versions -- instead of using TRANSFORM structures it now normally makes use of World Coordinate System (WCS) components of the images. For much of the time, and in particular to do the things which CCDPACK used to do, it is not necessary to be aware of this, but it opens the way to some new functionality. Discussion specific to the old methods can be found in appendix §.