In optical measurement of pipe threads at the manufacturing plant level, achieving uncertainty of within 1 μm was a problem because the object position and shape are not all the same. The authors propose a measurement method by designing an optical specification to suppress the diffraction effect and adopting optical shading/aberration correction and subpixel edge-detection processing. In this method (Quadruplet-Camera system), 4 CCD line cameras are positioned opposite to 4 parallel light sources in order to measure four points of the pipe circumference, and thread images are acquired by moving the Quadruplet-Camera system in the pipe axial direction. In the usual optical approach, the numerical aperture (NA) of the lens is increased to improve optical resolution. However, as NA increases, diffracted rays which form a diffraction pattern are observed in the edge area of a cylindrical object, and this causes measurement uncertainty. Increasing NA also results in a narrower depth of field (DOF), which causes instability in the measurement results at a manufacturing plant. The authors designed a compatible optical specification, and concluded that stability of measurement, which means eliminating the diffraction effect and securing a wide DOF, should take precedence over high optical resolution for application to a manufacturing plant, and adopted optical shading/aberration correction and subpixel edge-detection processing in order to compensate for lower optical resolution. In this manuscript, first, we explain the proposed method and confirmation experiments in the laboratory. We also explain a new optical measurement system based on the concept described above in a manufacturing plant to confirm the effectiveness of the method. We concluded that the measurement system has sufficient performance, i.e., uncertainty within 1 μm, for use as a practical system.
Optical surface inspection of steel mill products such as pipes, plates and slabs usually has the problem of overdetection,
which is caused by signals from harmless parts such as scale and surface texture. The authors propose a new inspection
technique based on the experience that most harmful defects on these products have a concave or convex shape, whereas
most harmless parts that might be overdetected have flat shapes. The proposed technique is called the ‘twin-illumination
and subtraction technique’.
In this technique, firstly, two images of the target area on a steel surface illuminated from the two sides are captured,
respectively. A subtraction image is then calculated from these images. Comparing the images illuminated from the
different sides, the images from concave or convex defects look different due to their different shadows, while images
from harmless flat parts look the same because illumination does not cause any shadow. As a result, two images with the
same appearances from a harmless part are canceled by subtraction, and two images with different appearances from a
concave or convex defect remain even after subtraction. Finally, it is possible to detect only concave or convex defects
without overdetecting flat patterns.
In this manuscript, first, we explain the proposed technique and confirmation experiments in the laboratory. We also
explain a new optical inspection system based on the concept described above and its application to moving hot pipes in
a steel manufacturing plant to prove the effectiveness of the technique. We concluded that the inspection system has
sufficient performance for use as a practical system.
The accuracy in the arrangement of grooved rolls for the finishing rolling mill is of large importance for the good
roundness of the bar steel product supplied to the precision machinery components such as the bearing of the high speed motor. Combining telecentric optics, silhouette image processing techniques, and statistical data processing allowed the development of the quantitative alignment guidance technique of the grooved rolls. The developed system demonstrated a high measuring accuracy and was seen to have practical use.