Object-Oriented Volume Segmentation

In this article we discuss three-dimensional image processing. Algorithms and data structures for this purpose are combined to form classes and objects in an object-oriented image analysis system. The major classes are volumes, octtrees, and image cubes. They provide reusable, problem-independent software components and hide implementation details. As an example, we show how data-driven volume segmentation of NMR-images can be accomplished using general assumptions about the image data. We point out how the classes can be integrated in an knowledge-based analysis system.