Image Processing

Image processing is any form of processing applied to an input image and whose output may be either the processed image or a set of parameters or features extracted from the image itself.

The processing applied to the image may run from primitive operations such as noise reduction, contrast enhancement, image sharpening, to higher level operations such as segmentation and classification of individual objects within the image.

In the former case, the output processed image is in a form more suitable for the particular needs, e.g. by increasing the contrast the image may look simply better or some obscured details may be brought out; another example of such processing is the compression, i.e. the application of techniques for reducing the storage/bandwidth required to save/transmit the image.

In the latter case, the output represents some attributes extracted from the image, e.g. edges, contours, objects, i.e. the output can be seen as a different description of the image. There is another area of image processing called image analysis, that performs the cognitive functions for understanding the image and its characteristics, for example the identification of a person from a face.

The fields of application of such image processing technologies span across science and industry, including medicine (e.g. detecting cancer in a magnetic resonance image); astronomy (e.g. calculating the size of a planet); materials/materials science (e.g. determining if a metal weld has cracks); machine vision (e.g. count items in a production line); security (detecting a person's eye color); robotics (e.g. recognizing an obstacle); and many other.

Right from its start, the IAPP research team is focusing on three main areas:

 cultural heritage, for example:

• virtual restoration for cracks and lacuna filling;
• multispectral digital imaging technique to achieve material localization and identification on painting surfaces;

medical applications, for example:

• analysis of angiographies for the automatic measuring of diameter and temporal pulsatility in blood vessels;
  

  industry, for example:

• parameter estimation for the automatic recognition of textile (cheratinic) fibres;
• view morphing techniques for building synthetic images between two images of an object taken from two different viewpoints.