Shortcuts in production-systems- A way to include clustering in structural pattern recognition

It is important that the semantics of production systems for image analysis are completely separated from the control. I. e., production application in different order or in parallel should not affect the result but, only the computational effort. In order to include also clustering, Hough-like accumulation and statistical models certain standard sub-systems are introduced. These can result in very high computational effort. In order to be applicable shortcut productions are defined for these sub-systems which have object sets of arbitrary finite size on their right hand side. For these productions the standard parsing algorithm must be modified - and the control independence only holds approximately and under certain conditions. The corresponding theory is elaborated and examples are given.