WO2003044505A1

WO2003044505A1 - Device for the detection of surface defects on cylinders

Info

Publication number: WO2003044505A1
Application number: PCT/ES2002/000553
Authority: WO
Inventors: Eusebio Bernabeu Martinez; Juan Carlos Martinez Anton; Luis Miguel Sanchez Brea; Philip Siegmann; Christoph Kraemer; Urs Peter Studer
Original assignee: Universidad Complutense De Madrid
Priority date: 2001-11-22
Filing date: 2002-11-22
Publication date: 2003-05-30
Also published as: ES2187381B2; WO2003044505B1; ES2187381A1; AU2002354198A1

Abstract

The invention relates to a device for the detection of surface defects on cylinders. More specifically, the invention relates to the production of a contactless optoelectronic device which is used for detecting, distinguishing and spatially locating surface defects or structures. Said device comprises three units: a) a light unit; b) a capture unit; and c) a signal processing unit. The invention is suitable for cylinders, cables or metal wires in general. In this specific case, the term surface structure refers to any topographic variation in relation to the cylindrical form.

Description

Device for detecting surface defects on cylinders

OBJECT OF THE INVENTION

The object of the invention is the realization of a device by means of which the surface defects present in cylinders are detected, especially in metallic threads, as well as their shape, spatial location and size with the cylinder in axial movement, that is, in line with production.

STATE OF THE TECHNIQUE

To date, some devices have been made for the determination of the surface quality of cylinders, among which we highlight those based on the illumination of the cylinder by a beam of light.

US 4 659 937, US 4 705 957, US 4 095 905 are known where a series of devices for detecting cylinder defects are presented.

These systems, however, have a number of problems. In the first place, the cylinder is not illuminated in a suitable way, since there are strong inhomogeneities in the lighting levels according to the area of the cylinder. Also, because monolithic photodetectors are often used, the information obtained from the surface structure of the cylinder is quite poor, since all knowledge of the surface of the metal cylinder is integrated into a single quality parameter.

At present, the manufacturing systems of the cylinders, specifically of metallic threads, demand greater performance and quantity of surface quality information. These problems until the Date have not been resolved. Through the device object of this patent we can: a) Distinguish certain types of defects (continuous structures, undulation and vibration marks). For continuous structures, we can: b) Give a parameter of quality of the surface state of the cylinder; c) Determine its spatial location and size. Both a), b) and c) are carried out on the production line ("in-line"). This has an interest in the field of metal wire manufacturing because, in addition to being able to determine the quality of the threads manufactured, it allows an analysis of the manufacturing processes, which can be used to improve those processes. By means of the presented device it will not be necessary to stop the production chain for its implementation.

The measuring device, object of this invention, is composed of three subsystems, as described in Figure 1:

• Lighting subsystem • Light pickup subsystem

• Signal processing subsystem

The most important features of the invention are:

• A lighting system that avoids areas insensitive to the detection of surface structures on the cylinder based on a single light source, although several sources could also be used.

• A detection system based on two-dimensional photodetector arrays, which collects spatial information on the surface state of the cylinders. • Two modes of operation:

A "visual" inspection mode (without processing), which allows a first distinction of the form or type of defect present in the cylinder: Continuous structures, undulation and vibration marks. With continuous structures we refer to defects with an axial length equal to or greater than the size of the inspection area (~ 2mm)). With undulation we mean a fluctuation of the cylinder axis of the Linear course of period greater than 2 times the size of the inspection area. With vibration marks we refer to a periodic fluctuation of the cylinder axis of less than 2 times the size of the inspection area. A signal processing mode that analyzes the distribution of light on the two-dimensional array of photodetectors and allows obtaining a two-dimensional map of the surface quality of the cylinder. Therefore, by means of this invention it is possible to detect the different sub-types of continuous defects that may exist on the cylinder, determine its position and estimate its importance.

• The device allows measurement in the production line and can be integrated into the production line without having to stop it.

DESCRIPTION OF THE INVENTION

Device for detecting surface defects on cylinders.

The invention relates to the realization of a contactless optoelectronic device for the detection and spatial location of structures or surface defects on cylinders.

Specifically, the system object of the invention is composed of three units: a) lighting unit, b) collection unit and c) signal processing unit.

a) By means of the lighting unit several beams of light are made to form equidistant angles between them on the surface of the cylinder to obtain a complete illumination of the contour thereof, and obliquely with respect to the axis of the cylinder. In this way, several light cones are generated which, by design parameters, do not overlap each other. b) The light scattered on the cylinder is collected by the pick-up unit, which deflects said light cones scattered by the cylinder on a two-dimensional array of photodetectors of CCD or CMOS technology. c) The signal collected by the array of photodetectors is processed in order to detect and obtain information about the location and characteristics of the surface structures on the cylinder.

In this way the device is composed of three subsystems, as described in Figure 1.

Lighting subsystem

A scheme of a possible lighting subsystem is shown in Figure 1. It is composed of a light emitter such as a laser, laser diode or Led (1), adjustable in intensity. These types of coherent or partially coherent sources produce a "speckle" or mottled due to said partial or total coherence. However, this effect disappears with the movement of the cylinder (3) and with the temporary integration that occurs in the collection process. The beam generated by this emitter affects a diffraction network (2) that divides the beam into several beams, according to the orders of the diffraction network. It will be attempted to have a diffraction network so that the light is distributed equally between the orders -1, 0 and 1. This can also be achieved by means of appropriate intensity filters.

Other types of inconsistent light sources, such as "white LEDs, halogen sources, etc., can also be used to illuminate the cylinder. These beams are directed by a series of mirrors (4, 5, 6) specially oriented so that they impact on the cylinder (3) obliquely from different sides (see Figure 2 for a section perpendicular to the axis of the cylinder showing said incidence) and on different points (x) although very close to the cylinder with different angles of incidence (α) (see Figure 3 to show the incidence of lightning), in order to illuminate a complete radial section (Figure 2) and prevent the exit light cylinder overlaps. The latter allows a correct detection of the surface structures on the cylinder.

If three equidistant beams are used, there will be zones ((13), Figure 2) that will be illuminated with two beams, so your information will be overlapping or duplicated. This is important, as we will see later, there are parts of the light reflected by the cylinder that is lost in the process of deflection of the conical beams reflected in the detection subsystem. Through this redundancy of information we ensure that we will not lose information from any of the illuminated areas of the cylinder. The optical system is designed in such a way that the areas whose information has been lost (due to the slit in the deflector mirror ((7), Figure 1) of the light collection subsystem) are those that have duplicate information, so that if one cone is lost, it remains in another.

Each incident beam is scattered across the cylinder, forming a cone of light (Figure 4), whose angle with respect to the axis of the cylinder (α) will be equal to the angle formed between the incident beam and the cylinder (Figure 3). As we have pointed out, to avoid an overlap of these cones in the detection plane (11) the angle (α) is different for each incident beam.

Detection subsystem

The light cones generated by the reflection on the cylinder reach a mirror ((7), Figure 1) that deflects said cones on the detection subsystem. (Figure 5). This mirror has a slit so that the cylinder can pass through it. The detection subsystem is formed by a suitable optical system consisting of (see Figure 1): lenses (8) and (10); mirror (7); mask (9) and by a two-dimensional array of CCD or CMOS technology detectors (11). The subsystem is specially designed so that the entire device can be inserted with the cylinder in motion, which is interesting for easy implementation in a production environment. The system works as follows: The beam dispersed by the cylinder forms a divergent cone that is collimated by a converging lens (8). These collimated cones affect masks (9) which are used to eliminate the maximum intensity in each cone (these maximums are the spot of the incident beams that do not reflect on the cylinder). This is done to avoid saturation of the arrays of photodiodes used. The masks are positioned in the plane of the waist of the laser beams. Subsequently, the light beams reach another converging lens (10) that reduces the size of the cones to a sufficient size to fit in the detector (two-dimensional array of photodetectors) (11). The intersection of the light cones with the detection screen can be seen in Figure 6. An image of the light distribution for a smooth cylinder and another with continuous surface structures is shown in Figure 7. Continuous structures cause annular fluctuations in intensity, while the flawless cylinder gives a "smooth" distribution of intensity.

An equivalent system for capturing the light cones reflected by the cylinder is shown in Figure 10 where the collimator lens (8) has been placed before the deflector mirror (7). In this case both with slit so that they pass through the cylinder. The advantages are: a) It will protect the mirror from dirt. It will be easier and less critical to clean the lens than the mirror, b) Reduce the loss of information due to the slits, by reflecting the 3 cones in the mirror with a larger radius, c) It will allow increasing the angles of incidence which it will improve the tolerances in the guidance of the cylinder and will allow to reduce the dimensions in the direction of guidance of the cylinder.

Another equivalent system is shown in Figure 11, in this case the two lens system and the deflector mirror are replaced by a single parabolic mirror that directly projects the three conical beams reflected on the detector, also interposing the corresponding masks. Signal Processing Subsystem

The signal generated by the array of photodetectors is processed by the signal processing subsystem, which may be formed by a) a computer with signal acquisition card or b) by an electronic card designed exprocess, which would include a "Digital Signal Processor "or DSP. The purpose of said signal processing is to detect the presence of defects on the surface of the cylinder, as well as to locate its position and determine its size.

The purpose of the processing is to automatically detect the presence or appearance of surface defects (mainly continuous structures such as stretch marks and grooves), locate them and at the same time obtain a surface quality parameter that allows quantifying the degree of surface deterioration due to these defects. We restrict ourselves to the detection of continuous defects (sufficiently elongated axially) because the axial movement of the cylinder in the production line implies an integration in the detection of local defects. A defect will disperse the rays that affect it in directions other than the specular direction corresponding to its position on the smooth cylinder, that is, a minimum of the intensity dispersed in this specular direction will appear. According to the profile of the defect this dispersion of the rays that affect it will be different, but remaining in the reflection cone (if the defect is axially elongated). If the density of defects increases, consequently they appear more minimal in the intensity in the reflection cone, that is, the intensity profile of the cone is less homogeneous. There is therefore a correlation between the density of defects and the degree of non-homogeneity of the distribution of the reflected intensity profile (interferential phenomena due to the interaction between defects, and between defect and smooth surface, will contribute to very high frequency fluctuations and they will not be able to be resolved by the CCD). The surface quality parameter is going to be a measure of this "degree of homogeneity." The parts of the processing are shown in Figure 9. Said processing is described below:

• Capture and radial and angular sampling of the three cones (after defining the positions of the masks). We pass from polar coordinates of the pixels (x, y) to cylindrical coordinates (r, ψ).

• Obtaining a smoothed profile corresponding to the perfect cylindrical surface. Two steps:

• Obtaining a single annular profile (for each cone) per average radius. We pass from Z (r, ψ) to T _mess (Ψ). • Polynomial adjustment of order n of T _mess Q ¥) (for each cone):

We obtain ϊ _{po¡ n} ¥) - This intensity profile simulates that obtained by the corresponding smooth surface (n = 3, low order).

• Obtaining the surface quality parameter T _n . Defined of the form:

where Λ / _ψ is the annular sampling. T „takes values between 0 and 1, it will be the closer to 1 the smoother the surface of the cylinder. A quality parameter of each reflection cone is calculated and its average value is finally taken.

• Superficial reconstruction. We associate to each point on the surface of the cylinder (r, φ) the intensity corresponding to it by specular reflection if the cylinder were smooth

where the intensities now also have a radial variation of the intensity weighed against the measured radial variation of the measured intensity.

The areas where the surface inspection overlaps are eliminated, and the areas of the surface where the information has been quantified are quantified lost due to the recess in the deflector mirror and the masks. • Location of defects. Where there is a defect, the intensity reaches a minimum. We locate the possible defects by means of a thresholding:

< ^I pol.n - ^J pol, _n > ^Ec l- ( ² )

where c is the sensitivity (we take c = 1), and where the intensities are passed to the coordinates of the cylinder surface (r, φ).

DESCRIPTION OF THE DRAWINGS

The invention is illustrated with 11 figures representing the following:

Figure 1: Scheme of the prototype for inspection of the cylinder surface.

Figure 2: Cross section to the axis of the cylinder (3) where the areas illuminated by each incident beam (A), (B) and (C) are shown.

Figure 3: Scheme of one of the incident rays reflected in the cylinder (3). Figure 4: Diagram of the light cone that is formed after the reflection of the incident beam on the cylindrical surface.

Figure 5: Scheme of the detection subsystem showing the deflector mirror and the two lens system.

Figure 6: Diagram resulting from the intersection of the light cones with the detection screen. The areas where there is a lack of lighting (15) and (16) are due, respectively, to the plugging with the masks and the slit in the deflector mirror (7).

Figure 7: Image of the distribution of light collected by the detector (11) for a smooth cylinder and another with continuous surface structures. Figure 8: Map of the surface of the cylinder showing the location obtained from the surface defects for the threads of Figure 7.

Figure 9: Diagram of the parts of the image processing obtained by the system to determine the surface status information as shown in Figure 8.

Figure 10: View of one of the possible configurations of the prototype in which the system is inserted into a waterproof housing.

Figure 11: Scheme of another possible configuration using a parabolic deflector mirror (7 ') to project the conical beams onto the detector.

EMBODIMENT OF THE INVENTION

The invention is illustrated by the following example:

We begin with a general explanation of the optical system object of the present patent:

The industrial implementation of the optical system set forth in Figure 1, requires first of all a design that allows a complete inspection of the entire surface contour of the cylinder, for which at least two conical beams will be needed that come from the reflection on opposite sides of the cylinder, because each scattered conical beam carries the information of the corresponding illuminated surface. However, the fall in intensity corresponding to the rays that reflect near the edge of the cylinder, and the effect of the diffracted rays, make it impossible to recognize defects found in these areas of incline near the edge. Finally, at least three scattered conical beams are necessary to obtain "recognizable" information of the entire peripheral surface of the cylinder. In our system the cylinder will be illuminated with these three incident beams that are equidistant from each other (forming ~ 120 ° between them) such that each one reaches to illuminate the surface that does not become illuminated by the other two.

In Figure 1 we show the optical system, which allows a complete peripheral inspection of the surface of the cylinder, and where the three conical beams are diverted outside the axis of the cylinder for later capture with a CCD camera.

The three conical beams that are formed have the same axis as that of the cylinder (3) in their lighting zone (12), so they cannot be captured with a CCD (11) placed perpendicular to the axis (unless the photodetector had a hole through which it could pass through the thread). The three conical beams must be projected outside the axis of the thread.

Through a diffraction network (2) on which we hit with a laser beam (1) we generated three diffracted coplanar beams (orders -1, 0, + 1), (see Figure 1). Each of these three beams is reflected in its mirror (mirrors (3), (4) and (5)) that gives them the right direction to: a) Obtain complete illumination around the cylinder, and b) get the three beams Scattered conics can be caught by a deflector mirror (7) without overlapping, which is achieved by slightly varying the angle of incidence on the cylinder for each beam, in the case that we keep the cut-off point with the same axis for all three ( what we will initially maintain). This deflecting mirror (7) is located on the axis of the cylinder forming 45 ° with it. To make this possible, the deflector mirror has a cut or groove through which the cylinder passes (and that will allow an implementation of the entire device in the production line without need to thread the cylinder through the mirror (7)). The central incident beam (of order 0), directed by the mirror (4), will form the central dispersed cone and the part that does not reflect in the cylinder will pass through the mirror cut (7). The other two beams (order -1 and +1) directed by the mirrors (5) and (6), will affect the cylinder from other sides forming the inner and outer reflection cones. Due to the slit in the mirror (7) each scattered cone will lose a small piece of arc.

The three conical beams scattered outside the axis of the cylinder collide with the lens (8) whose focal point coincides with the point of convergence of the cones (~ lighted area of the wire (12)). The conical beams are now cylindrical (not so the thickness of the beams that follow another convergence) until they reach the second converging lens (10). In this path they cross a series of masks (9) that eliminate the interferometric part and the part of the incident beams (of order -1 and +1) that have not reflected in the cylinder (spot). The second converging lens (10) will project the three cones on the CCD camera (11) without the interferometric part (14). We will call these cones masked reflection cones. The information of the region of the cylinder that reflects towards the slit of the deflector mirror (7) will also be lost, but with an appropriate adjustment of the mirrors (5) and (6) it is possible to recover the information lost by the slit in the mirror ( 7) of a cone with the information of this region of the surface that the other cone carries with it.

In the following figures we show in more detail the steps followed in the collection of the signal by the optical system of Figure 1: The areas illuminated by each incident beam are shown in Figure 2 with (A), (B) and (C), as well as areas that are illuminated by two beams (13). The lighting of a section of the cylinder is complete, so no information about surface defects is lost. The angle formed by the scattered conical beam with the axis of the cylinder is the same as that formed by the incident beam. A diagram of one of the incident rays reflected in the cylinder (3) is shown in Figure 3. It is shown that the angle (α) between the cone (to which the reflected beam belongs) and the cylinder is the same as between the incident beam and the cylinder. The point of reflection is (x).

In Figure 4 we show a scheme of how one of the incident beams generates a cone of light. Where (α) is the angle of the incident beam relative to the axis of the cylinder, (12) is the lighting zone of the cylinder, and (14) is the cut of the reflection cone with the plane perpendicular to the axis of the cylinder. The scheme of the detection subsystem is taught in more detail in Figure 5, where it is shown how the optical design collimates the rays of the light cone and is filtered through masks to eliminate the maximum intensity in the cone. With (3) we designate the cylinder, with (12) the lighting zone of the cylinder, with (7) the deflector mirror with a slit to be able to introduce the cylinder and inclined to remove the three light cones out of the cylinder axis it reflects, with (8) the collimating lens, with (9) the masks used to plug the areas of maximum intensity that can saturate the two-dimensional array, with (10) the converging lens whose purpose is to project the rays on the detector (two-dimensional array of photodetectors), and with (14) the intersection of the light cones with the detector screen.

The image obtained by the configuration of Figure 1 of the illuminated zone (12) of the cylinder, is schematized in Figure 6, where the dark part is the illuminated areas. The circular sections are not closed. The areas where lighting is lacking (15) and (16) are due respectively to the plugging with the mask and the slit in the deflector mirror (7) where "part" of the reflected conical beams is "lost."

We now show an example of the processing of the signal collected by the optical system of Figure 1:

Figure 7 shows example images obtained with the CCD camera of the optical system of Figure 1 and by a conventional optical microscope of metallic wires with different surface finishes. Fluctuations in the intensity detected carry information about the location and dimensions of surface structures. Figure 7. a) Image of the surface of a 55 mm diameter "smooth" metallic wire taken by a 20: 1 magnification optical microscope. b) Image obtained by the system object of the invention of the same area of the thread as in Figure 7a. c) Image of the surface of a "defective" metallic wire of 120 microns in diameter taken by an optical magnification microscope: 20: 1. d) Image obtained by the system object of the invention in the same area of the metallic wire of Figure 7c.

After an analysis as described above, some "pseudo-images" shown in Figure 8 are obtained for these two cylinders shown in Figure 7, where the defects found by an automatic and appropriate thresholding method are remarked and located. intensity profile of the reflection cones. In this figure, the quality parameters 7 _{3 are also given} , highlighting that both cylinders can be completely differentiated, and the percentage (annular) of the inspected surface S. Figure 8a) corresponds to the thread "with defects" of Figure 7c and Figure 8b) corresponds to the "defect-free" thread of Figure 7a.

The summary diagram of the steps followed in the processing of the image collected by the detector of the optical inspection system of the cylinder surface to obtain the parameter of surface quality and location of defects is shown in Figure 9.

Scheme of other possible configurations of the optical system object of the present patent:

A prototype scheme for inspection of the cylinder surface is shown in Figure 10. With (1) we designate the light source (s), with (4,5,6) the addressing mirrors, with (3) the cylinder, with (12) the inspection area of the cylinder, with (16,17) the protective windows, with (8) the collimating lens, with (7) the deflector mirror, with (9) the masks, with (10) the collimating lens and with (11) the detector. Figure 10 b) (side view) shows the detection subsystem where the collimating lens (8) has also been placed in front of the deflector mirror (8) so that it passes through the cylinder.

Figure 11 shows the scheme of an alternative detection subsystem that replaces the deflector mirror (7) of Figure 1 and 5 with another parabolic (7 ') that allows the three reflection cones to be projected directly on the detector without the need for lens system (8) and (10).

The scheme in Figure 10 shows an example of a possible configuration for the system object of the invention. In this example it has been sought that: a) The system can be introduced into a housing in which the The cylinder is always external to the system so that it can be inserted "¡n-line" in the production chains of the cylinder, b) The optical parts are protected from external agents such as dust, liquids and other types of dirt. It should be remembered that metal wire production systems are very aggressive environments for optical designs (optics can be fogged with vapors and splashes of lubricants). According to the example system of Figure 10 all optical components are inserted in a sealed housing where there is a flat window (17) for easy cleaning and several holes (also with a window) (16) where the light beams that illuminate the cylinder

Claims

1. An optoelectronic system for the detection of surface structures on cylinders consisting of: An optical lighting system

An optical system for detecting the rays scattered throughout the cylinder. A signal processing system detected by the array of photodetectors

2. Optoelectronic system for the detection of surface structures on cylinders, according to claim 1, characterized in that the optical lighting system is based on a monochromatic light source (LED, laser or laser diode, etc.) whose beam is divided into several beams by means of a diffraction net, which are directed so that they impact in an equidistant way and with angles of incidence with respect to the axis of the cylinder slightly different.

3. An opto-electronic system for the detection of surface structures on cylinders, according to claim 2, characterized in that the lighting system consists of several light sources, both monochromatic and polychromatic (white light diodes, sources, halogen, etc.) , with adjustable intensity.

4. An optoelectronic system for the detection of surface structures on cylinders, according to claim 1, characterized in that the detection system consists of a deflector mirror with a slit through which the cylinder crosses and centered and inclined with respect to the axis thereof with the in order to take out the beam from the cylinder axis so that they are subsequently captured by the detector.

5. An optoelectronic system for the detection of surface structures on cylinders, according to claims 1 and 4, characterized in that the detection system is composed of: collimating lens placed before the deflector mirror which in turn is placed before the converging lens.

6. An optoelectronic system for the detection of surface structures on cylinders, according to claims 1 and 4, characterized in that the Detection system comprises a: deflector mirror placed before the collimator lens which in turn is converged lens.

7. An optoelectronic system for the detection of surface structures on cylinders, according to claims 1 and 4 characterized in that the

The detection system consists of a parabolic deflector mirror that makes the conical beams converge directly on the detector.

8. An optoelectronic system for the detection of surface structures on cylinders, according to the preceding claims, wherein the detection system consists of a two-dimensional array of photodetector elements

10 in CCD or CMOS technology.

9. An optoelectronic system for the detection of surface structures on cylinders, according to claim 8, wherein the detection system consists of an array of photodetectors in annular arrangement.

10. An optoelectronic system for the detection of surface structures on cylinders, according to claim 1, characterized in that the processing for obtaining information on the surface structures will be carried out in a computer with a signal acquisition card or in an electronic card whose main element be a DSP (Digital Signal Processor). 11. An opto-electronic system for the detection of surface structures on cylinders, according to claim 10, wherein the signal processing allows to calculate a quality parameter through which the structures are quantified and located "¡n-line" continuous on the surface of the cylinders.

25