Methods for digital image processing

In this tutorial, the fundamentals of digital image processing from a historical perspective will be discussed, placing particular emphasis on digital filters and segmentation methods Image-analysis topics will include techniques for the geometric and densitometric description of the structures identified in an image. Such techniques are essential for image understanding and classification purposes. In particular, an overview of model-based object identification, model instantiation, feature extraction and image matching methods will be provided. Problems and methodologies regarding the optimal symbolic representation of images (with respect to the source data and the application domain) and consistency with a priori knowledge will also be presented. The various methods will be explained through numerous examples. Some insight into the theory will be provided, so as to facilitate further reading. An outline of the MPEG7 standard for image-feature encoding will also be presented. Finally, we will give a brief account of recent research activities carried out at our institute ISTI-CNR