Real-time flatness inspection of rolled products based on optical laser triangulation and three-dimensional surface reconstruction

Abstract

Abstract. Flatness is a major geometrical feature of rolled products specified by both production and quality needs. Real-

An improved 3D imaging system for dimensional quality inspection of rolled products in the metal industry

“…This system is based on optical triangulation using a single-line pattern which is projected onto the surface of the product under inspection to reconstruct its surface. A complete description of the laboratory prototype and the system installed in a conditioning mill of ArcelorMittal in Asturias (Spain) can be found in [17]. After four years of usage of this system we performed a thorough analysis to improve its performance and to adapt it to the ever-growing quality requirements in the metal industry.…”

“…Therefore, measuring the length of a longitudinal fiber of a rolled product exhibiting flatness defects involves computing the length of the waves of its surface. In our previous works, [17,18], we described in depth how to measure the flatness of rolled products based on the information provided by an active triangulation system.…”

“…The calibration of the proposed system follows the non-linear camera model proposed in [44] that maps 3D world coordinates ðx w ; y w ; z w Þ to pixel coordinates in the 2D camera sensor (r, c) by means of several transformations. The calibration procedure following this model is described in depth in [17].…”

“…In previous works we have developed a laboratory and an industrial 3D imaging prototype for surface reconstruction of long, flat-rolled products in the metal industry based on optical triangulation using a single-line laser pattern [17]. The prototype was designed as part of a real-time quality control system and provides both flatness and width measurements of the rolled products under inspection [18].…”

Software architecture for time-constrained machine vision applications

“…2 Requirements, Attributes, and Design Principles Based on our experience designing and implementing machine vision applications, [14][15][16][17] the most important nonfunctional requirements of a real-time image processing architecture are the following:…”