A method of determining an infrared sensor for automated inspection of hot metal
sheet is described. The sensor locates voids, inclusions, and surface defects
characterized by a variation from background radiant flux. The sensor can be
coupled with a commercially available processing system, developed for visible
wavelength line scan cameras, to automatically track, classify and report product
flaws in real time.
Theory: For hot metals, emissivity is strongly dependent upon surface conditions.
Any scratch on the material will radiatemore strongly than smooth regions due
to multiple reflections between the sides of the indentation. Changes in
emissivity are also observed where surface blemishes occur, with darker regions
emitting more than lighter areas. In addition, any void or inclusion within the
metal may result in conduction properties different from metal of homogeneous
composition. If these flaws are near enough to the surface, they will alter the
radiance in localized regions.
Approach: To detect signal modulations of a few percent in areas as small as a
scratch, a high resolution infrared scanner is required. In some hot metal
sheet production facilities, the web width can range from three to seven feet
while the line speed can exceed 1,000 ft/mm. In order to achieve 100% coverage
under these conditions, the inspection system requires both a fast response
sensor and rapid signal processing capability. Currently, there exists at least
one commercially available image processing system capable of tracking and
reporting surface anomalies in real time. Here we present a method for
specifying a sensor based on properties of the target material, environmental
considerations, and commercially available infrared detectors.