Enhancing the accuracy of forging measurement using silhouettes in images

In this study, a new passive camera system for heavy cylindrical forging measurement, based on silhouettes in images, is developed. New methods for making such a system more resistant to the negative effects of the industrial environment have been proposed. This includes weighted edge filtering based on complementary information about the edge quality in an image. The recorded measurement median errors were ±0.12 mm and ±0.14 mm. Moreover, the 95% confidence intervals were ±0.5 mm and ±1 mm for the forging axis and diameter measurements, respectively. Both results were achieved in a measurement volume of 6 × 6 × 2 m during the measurement of glowing hot forgings in industrial conditions. The results surpass those of the state-of-the-art method, mainly in the case of axis straightness measurement, by approximately 50%. The measurement is fast, and it provides feedback about the axis straightness for its subsequent correction.