Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 1994 · 37 citations · 0 references
ComputingComputer ScienceFeature EngineeringKnowledge RepresentationMachine LearningDiagnosisExpert SystemsFault DetectionFault DiagnosisPattern EvaluationPattern RecognitionAutomatic Fault Detection SystemsDecision RulesFault ClassificationFringe AnalysisHierarchical ClassificationKnowledge-based Fault DetectionKnowledge-based Fault DiagnosisSystem Fault Diagnosis
Automatic pattern recognition methods combined with a knowledge based approach are used for the detection and classification of fault indicating patterns. For this purpose the complex fringe pattern is reduced to a line pattern (fringe skeleton) and this line pattern is approximated by vectors. The new pattern representation is transformed in a data list preserving the topology and metrics of the original pattern. Based on characteristic symptoms within the pattern features for the description of different pattern classes are derived. The knowledge for the hierarchical classification procedure embedded in decision rules for the knowledge assisted system was derived from model based simulation of fringe patterns and preparation of test objects with different flaws.