Processing system for detection and the classification of flaws on metallic surfaces

Claim 1

In an automatic surface scanning method of the type in which surfaces have been subjected to at least two manufacturing steps, said method utilizing advancing a surface past a source of electromagnetic radiation, directing said electromagnetic radiation toward said surface, sensing reflected radiation from said surface with electromagnetic-radiation sensing means, and effectively scanning at least one of said source and said radiation sensing means across said surface generally transverse to the direction in which said surface is advanced to produce a series of serial scan signals, an improvement utilizing defect classification, said improvement comprising: a. using a first radiation sensor and a second sensor, each sensor generating a signal which is a function of the amount of radiation received; b. adjusting the relative position of said radiation source and said sensors so that an essentially defect-free surface reflects radiation principally to said first sensor and reflects a measured amount of radiation to said second sensor; c. generating a first sensor, position-related, average signal; d. comparing said first sensor signal to said average signal and generating a first error signal if said first sensor signal exceeds said average signal by a first predetermined amount, and generating a second error signal if said first sensor signal is less than said average signal by a second predetermined amount; e. comparing said second radiation sensor signal to said first sensor average signal and generating a third error signal when said second signal exceeds a first predetermined fraction of said average signal and generating a fourth error signal when said second signal exceeds a second predetermined fraction of said average signal, both said first and said second predetermined fractions being greater than the ratio of said measured amount of radiation received by said second sensor from said essentially defect-free surface to said first sensor average signal and both said fractions being less than one, and said second predetermined fraction being larger than said first predetermined fraction; f. detecting the essentially simultaneous occurrence of at least two preselected conditions involving at least two of said error signals to provide defect classification signals; g. utilizing said classification signals to indicate rejection of said surface and to indicate the defect classification; and h. utilizing at least one of said classification signals to indicate to at least one of said at least two manufacturing steps that there has been an occurrence of a classification defect, whereby process control can be implemented to minimize the future occurrences of said classification defect.