Experience in Integrating Image Processing Programs

摘要

This paper describes experience we have gained in using program supervision techniques for integration and control of image processing programs. The role of these knowledge-based techniques is to handle high level knowledge about the use of programs together with their effective execution and control. First, the main concepts of program supervision techniques are introduced. Then we present three major classes of application domains where we have applied program supervision techniques: road obstacle detection, medical imaging and astronomy. From these experiments, we conclude that program supervision techniques are good candidates to perform a semantical integration of image processing programs. These techniques are independent of any particular application domain or image processing programs. We propose a general architecture for such systems, which can be slightly tuned for specific purposes. For code designers, semantical integration implies to think from the beginning about how their image processing programs can be integrated and controlled.