High-speed defect detection method for color printed matter

The authors present a novel defect-detection method based on image processing technologies, aimed at automatic inspection of color-printed matter. In this method, a reference pattern containing allowable ranges is expressed in an index space which is a three-dimensional flag table constructed from a gray-level axis and two planar coordinate axes. Each pixel of the test image is inspected by taking its three-dimensional address in the index space and then referring to the corresponding flag in the index space. This method allows images to be inspected a high speed using human-like judgment criteria. The validity of the method is demonstrated by experiments using a prototype for inspecting the printed surface of prepaid cards. >