EXTRACTION OF RULES FROM NATURAL OBJECTS FOR AUTOMATED MECHANICAL PROCESSING

In process applications, fast and accurate extraction of complex infor mation from an object for the purpose of mechanical processing of that object, is often required. In this paper, a general rule-based approa ch is developed using a database of measurable geometric ''features'' and associated complex information. The rules relate the features to t he complex processing information. During the on-line processing, the object features are measured and passed into the rule base. The output from the rule base is the complex information that is needed to proce ss the object. A methodology is developed to generate probabilistic ru les for the rule base using multivariate probability densities. A know ledge integration scheme is also developed which combines statistical knowledge with expert knowledge in order to improve the reliability an d efficiency of information extraction. The rule generation methodolog y is implemented in a knowledge-based vision system for process inform ation recognition. As an illustrative example, the problem of efficien t head removal in an automated salmon processing plant is considered.