Software engineering to augment image processing tools with a priori knowledge about applications

Deals with a specific approach currently tested in the authors´ laboratory, relative to the development of an user oriented interface dedicated to image processing (IP). The main concern is to integrate different classes of IP methods into a programming environment which allows various IP concepts to be handled. It is possible to describe an algorithm using associative mechanisms on abstract objects including inheritance of their properties from top to bottom and to store it into an IP database. Each treatment to be included in the programming environment gets a hierarchical functional representation; it is then defined as a graph linking some operators working on a set of basic IP objects and carrying IP properties. This kind of `standardization´ helps to classify, select, modify or create operators. This way any theoretical model implicity embedded in an IP algorithm could be described in terms of its properties, so the propose a semantic approach aiming to capture the `a priori´ knowledge about, for instance, applications in taking into account the hierarchical links between models and patterns, patterns and operator properties then properties and access criteria