Scale-space coding and its perceptual relevance

In our research on image coding algorithms we have adopted the following starting points. First, processing by coding algorithms should as closely as possible match what we know about the human visual system. Second, due to the lack of acceptable objective criteria, proper evaluatioll of coding algorithms, as well as parameter settings, require perceptual ex periments. In this paper we expand the so-called scale-space model for image coding and demonstrate its relevance to visual perception. In the scale-space model an image is passed through Gaussian filters of decreasing bandwidth. The variation between succeedingly filtered responses is very systematic, so that little information is needed to pass between them. Starting from a low bandwidth version of the original image, we make a prediction for a higher bandwidth version. Only the prediction errors need be transmitted to recover this higher resolution picture. The process is repeated at a number of levels (called scales) in order to arrive at the original image. For data-reduction purposes, several approximations or t.hese prediction errors are studied. Results of preliminary perceptual experiments, conducted in order to make an optimum choice for some of the coding parameters, are also discussed.