2. Choice of attributes

The choice of attributes used to guide the registration process is crucial. It is largely conditioned by the nature of the images to be processed. Two approaches can be distinguished: geometric and iconic. Geometric methods involve extracting, either manually or automatically, scattered geometric primitives from the image (points, curves or surfaces) and matching them. Iconic approaches use the dense set of information carried by the image's gray levels, either directly by comparing intensities, or by comparing image values after processing (e.g. by calculating differential image characteristics or considering the Fourier transform). Hybrid methods combining these different approaches have also been proposed.