Overview

What is Hermes ?

Hermes is a human-computer interface that allows users to formulate image processing application problems. This problem has to be solve by an image processing application. The formulation is composed of two main parts :

• objectives specified by a list of image processing tasks.
• an image class defined by a list of invariant features (given by numeric and symbolic descriptors).

Who is concerned by Hermes ?

Users inexperienced in the image processing field who needs to transform automatically a set of images are concerned. These users have to be experts in the application domain to be able to:

• express their objectives of image transformations.
• define the acquisition chain effects on the resulting image (noise, defects, ...).
• describe the visualized scene content (by the business concepts of the user's domain)
• evaluate the transformed images which will be used for further processing (analysis, interpretation, couting, measures, ..) named post-processing tasks.

How does Hermes work ?

Hermes supervises the interaction with the user and translates the user's description (in terms of his domain) into terms of image processing. Then it constructs a request which is the formulation of the problem defining the needed application and sends it to a planification system for image processing tasks (Borg). This planification system gives the image processing software that suits the user's requirements.

The interaction model which has been designed allows the user to correct his formulation. He gets back transformed images, is able to evaluate them and he can modify his formulation and asks for a new solution.