On-line measurement of the straightness of seamless steel pipes using machine vision technique
Introduction
Straightness is of critical importance for guaranteeing the quality of the screw threads machined at the ends seamless steel pipes. In order to monitor the product quality in product lines, 100% on-line measurement is required. Here, the straightness and diameter of seamless steel pipes over 10 meter long need to be inspected [1], with required straightness tolerances of <3 mm over the whole length. So far, no satisfactory method has been in practical use to solve the on-line automatic measurement of the straightness. Therefore, the straightness evaluation has to be accomplished manually by operators. This not only results in low measurement speed, but also makes the production quantity subject to human errors.
Over the years, different measurement methods for the related applications have been explored. These include the use of alignment telescope, jig transits, optical levels, the engineering theodolite and so on [2], [3]. However, none of these are directly suitable for the 100% on-line measurement of the spatial straightness of large workpieces. Junichi et al. [4] used laser beam to obtain the straightness by moving target or mirror along the measured line or the measured central axis. However, there are two detrimental factors for the laser datum stability. One is the air flow in the beam path. The effect of air flow can be reduced by putting a comer (tube) around the beam path for small measurement ranges. However, for large objects, the efficiency of this method is not apparent. The other problem is the drift of the beam which is caused by “beam dancing” due mainly to the thermal deformation of the parts of the laser. The low drift of 3 nm over the working distance of about 200 mm can be expected by stabilization using water cooling for the laser tube. However, this stability is not sufficient for the measurement of straightness of longer parts such as the working distance up to 1 m [4]. In order to implement the straightness measurement method for a large object with multi-point displacements, Sasaki et al. [5] also used laser beam as the straightness reference. The only difference from the previously mentioned one is that the detector consists of four-cell type photodiodes inside a Si micromesh structure. The direction or the wave front of the reference laser beam is not disturbed, as sensor absorbs only some of the incident photons and transmits the remainder down stream. However, this method is still subject to the influence of environmental factors. Furthermore, it cannot be used directly for non-contact measurements. Parallel white light beam method was also attempted [6]. This method is not subject to the above detrimental effects, but like the Junichi’s method, it cannot implement the non-contact measurement, since during the measurement, the moved target or reflection mirror must be used. Herrmannsfeldt et al. [7] developed a zone plate method for straightness measurement. In this method, a system of large rectangular Fresnel lenses was used in the laser alignment system for the SLAC 2-mile accelerator. The alignment system consists of a He–Ne laser light source, a photoelectric detector, and the lenses, one of which is located at each of the 297 points which are to be aligned. Each lens has the proper focal length to focus the laser to a point image at the detector. This method suffers from the same drawbacks as above. These methods can only be used for off-line spot check of large object’s straightness. They are not suitable for 100% on-line non-contact automatic measurement of the straightness required here.
With the advancement of machine vision technology, visual gauging or inspection technique based on off-the-shelf charge coupled device (CCD), semiconductor laser, image grabber and image processing device, is becoming a an alternative for 100% on-line measurement in industrial applications. This method offers the desired properties of being non-contact, with proper precision, high speed and easy to operate. In this paper, we develop a laser visual alignment measuring technique to implement 100% on-line and non-contract measurement of straightness of large-scale object for the inspection of seamless steel pipe’s straightness.
In Section 2, we introduce the fundamentals of the visual alignment system and its mathematic modeling. Then we describe in Section 3 how to obtain the coordinates of the centers of elliptical arc via the visual sensing. In Section 4, the calibration of the measurement system is given. Finally, in Section 5, this novel method is verified by experiments in the straightness measurement of a seamless steel pipe with 1500 mm length.
Section snippets
Measuring principle and system modeling
The measurement of straightness refers to that of the 3D coordinates of the points in the central axes of an object [8]. Therefore, as long as there are sufficient numbers of measured points along the axes of a workpiece, its straightness can be estimated. The key problem in the measurement is how to measure the 3D coordinates of the points along the measured line. The method we used to measure the straightness of the axis line of an external cylinder is illustrated in Fig. 1. Several laser
Computation of the elliptic arc center
In practical measurement, the digital image of elliptic arc as shown in Fig. 3 can be acquired after the image of ellipse arc formed by the intersection between the light plane and the cylinder is sampled by image grabber and computer. Then each point’s coordinate of the elliptic arc in the frame memory can be computed through image processing. Suppose the coordinate of elliptic arc center, which is the intersection between light plane and axis, is c(xw, yw, zw) in the world coordinate system,
Calibration of the alignment system
From , , we know that there are 24 unknowns in the measured model. They are the effective focal length of the camera f, uncertainty scale factor sx, the coordinates of the optical center O in the image coordinate system (U0, V0), the coefficients of radial distortion k1 and k2, the rotation vector (r2,r5,r8)T and translation vector (tx,ty,tz)T between sensor coordinate system and camera coordinate system, and the rotation vector (R1,R4,R7)T, (R2,R5,R8)T and translation vector (Tx,Ty,Tz)T
Experimental results
In order to test the proposed method and obtain the measurement precision of the visual alignment measurement technique presented previously, we built a set-up of visual alignment experiment system as shown in Fig. 6. Using three visual sensor groups, each of the group consisting of two sensors arranged at the two sides. The camera used here is Mintron MTV-368P. The center to center distance between two adjacent sensor elements in X-direction dx=0.980 μm, and in Y-direction dy=0.6357 μm. The
Conclusion
The visual alignment system presented above proved to be successful in the experiments. In practice, in order to obtain high measurement precision, no less than three visual sensor groups along the axes to be measured should to be used. With more measurement points used, a higher measurement precision is achievable when the pipe straightness is evaluated. For the 1500 mm long seamless steel pipe, more than five sensor groups are advisable. From the result as shown in Table 2, we can observe that
References (14)
Further five-point fit ellipse fitting
Graph. Models Image Proc.
(1999)- et al.
From the Hough transform to a new approach for the detection and approximation of elliptical arcs
Graph. Models Image Proc.
(1998) - R. Lu, Study on the Visual Alignment Technique for On-line Measurement, The Research Report of Postdoctoral Research,...
- D. Vernon, Machine Vision — Automated Visual Inspection and Robot Vision, Cambridge University Press, Cambridge,...
- F.T. Farago, M.A. Curtis, Handbook of Dimensional Measurement, 3rd Edition, Industrial Press Inc.,...
- et al.
Straightness measurement using laser beam straight datum
SPIE Int. Soc. Opt. Eng.
(1995) - et al.
Transmission-type position sensor for the straightness measurement of a large structure
J. Micromech. Microeng.
(1999)
Cited by (67)
Development of a novel multi-line laser triangulation scanning system based on the rotary diffraction grating
2024, Measurement: Journal of the International Measurement ConfederationCalibration of line-structured light vision sensor based on free-placed single cylindrical target
2022, Optics and Lasers in EngineeringResearch on the on-line dimensional accuracy measurement method of conical spun workpieces based on machine vision technology
2019, Measurement: Journal of the International Measurement ConfederationCitation Excerpt :Li et al. [10] proposed a new image edge detection method and measured the straightness of the steel tube successfully based on machine vision technology. Lu et al. [11] developed a set of straightness measurement system for 1500 mm long seamless steel tube based on machine vision, which has high accuracy and stability. Sun et al. [12] proposed a machine vision method for on-line measurement of the cross section diameter and straightness of seamless steel pipe by placing a pair of line structure laser sensors on each sample section of the pipe.
High-precision diameter detector and three-dimensional reconstruction method for oil and gas pipelines
2018, Journal of Petroleum Science and EngineeringDeflection detection and curve fitting in three-roll continuous straightening process for LSAW pipes
2018, Journal of Materials Processing TechnologyVirtual laser vision sensor environment assessment for surface profiling applications
2018, Measurement: Journal of the International Measurement ConfederationCitation Excerpt :If not met, they can adversely affect the critical buckling load a pipe can tolerate, and the quality of welds [1,2]. Different measurement techniques were developed over the years, which ultimately led to the design of a complete real-time vision measurement approach to assess the straightness of large seamless pipes [1]. Since the measurement technique did not utilize advantages offered through the manufacturing process, the procedure required numerous triangulation sensors [2].