An opto-digital interferogram evaluation system for fault detection in quality control is presented. The paper concentrates on the optical preprocessing that serves data reduction and feature extraction. Various approaches are discussed and the chosen technique, a time-varying filter process, is treated in detail. Results produced with an experimental setup are shown.
Interferogram based fault detection is essentially a problem of pattern recognition: Typical fringe configurations indicating mechanical defects must be detected. For thiis purpose, data reduction and feature extraction are necessary in order to compute the final decision. The local grating-like structure of interferograms suggest their description in terms of line spacing and orientation. To extract and evaluate these features, a hybrid opto-digital processor is introduced.
In non-destructive testing, interferometric methods play an increasing rôle, especially as they are entering the field of quality control in mass production. An evaluation system for interferograms, designed for such applications, has to perform a typical pattern recognition task: Faults, e.g. in machinery, resulting in characteristic interference fringes, must be detected, thus leading to the desired “go” or “no go” statement.