Post date: 24-feb-2012 9.34.57
Riprendo il testo di Rioux (è possibile scaricare l'articolo)
5.4 Anthropometric modelling
For the future it is anticipated that the third level of compression will be to use as much knowledge as possible in an anthropometric model in order to extract, from the combination of the scanned data and the a priori knowledge, a small number of parameters that would allow precise reconstruction of the subject geometry. It is difficult to predict the number of parameters that would be required, but experience indicates that we can set the expected lower and higher bounds can be set to a few hundred and a few thousand, respectively (-10 kB?). Such a model-based approach could also include skin reflectance properties for colour diffusion.
If one assumes a lo-kB file, a CD-ROM could store the measurements of more than 50000 subjects. On the other hand, this approach will be limited in terms of compressing fingerprint patterns. The anthropometric model will very efficiently compress body structure and shapes that are generic to the human form, such as the elements of the face and their relative positions, or ears, for which large physical dimensions exist, but in terms of structure are very similar from one subject to another. This is not the case with fingerprints. Although the ridge-based structure is common to everyone, the complexity of the pattern and, most of all, its variability in structured pattern make it very difficult to model.
Il "modello ideale" dovrebbe essere in grado di contenere il massimo dell'informazione sulla forma e struttura umana in generale, mentre i parametri antropometrici avrebbero lo scopo di individuare la forma del particolare individuo.