3 Software Engineering for Image Processing and Analysis

3.1 Introduction Connguring or programming image processing systems is a time-consuming task which requires specialized knowledge on the eeects of image processing algorithms as well as knowledge about the implementation and interfaces. Clearly, software engineering is required for the application programmer of image processing systems. But even people who do not implement their applications themselves occasionally have to face software engineering problems. Several commercial or free packages for image processing exist providing routines which can be plugged together to more complex operations. This does not solve software engineering problems; it rather shifts the basic building blocks to a higher level. The larger the application gets which uses such libraries, the higher is the importance of well-structured software. The major problem in design of general imaging systems is that on the one hand highly run-time eecient code and low-level access to hardware is required, and that on the other hand a general and platform-independent implementation is desired which provides all data types and functions also for at least intermediate-level processing, such as results of segmentation. Software re-use is crucial in any large system; well-documented packages should be usable even across applications. If every programmer is allowed to reprogram existing code, soon several pieces of code will be scattered around in the system which serve the same purpose. Today's software engineering is closely coupled with the ideas of object-orientation. Theoretically and practically, object-oriented programming gains high attention. Object-oriented programming can help simplifying code re-use; if applied properly, it uniies