Flatness monitor

Abstract

A system method of monitoring flatness of a material. A beam of coherent light is projected on to a surface of the material to provide a line on the surface. An image of the line on the surface is obtained, and the image is used to determine a deviation of the line from a predetermined configuration. The deviation corresponds to an irregular pattern or unevenness of the material.

Description

This application claims priority from U.S. Patent Application Ser. No. 60/588,355 filed on Jul. 16, 2004.

FIELD OF THE INVENTION

The present invention relates to a method and apparatus for monitoring the flatness of a material.

BACKGROUND OF THE INVENTION

The flatness of a material is a paramount importance in certain applications. For example, in the steel industry, the flatness of a thin steel strip is critical to the effectiveness of subsequent processing. For example, such a strip of steel might be used in the manufacturer of cans, and the integrity of the tin coating process is dependent upon the flatness of the feed stock.

Whilst the flatness of the body of a steel strip can be maintained, the edges of such a strip are prone to deviate during the production of the strip and produce what is known as a “rippled” edge. This rippling, if beyond certain limits, can adversely impact upon the subsequent processing of the strip, and lead to failure of the can.

The flatness of the material is usually evaluated by obtaining a sample and measuring the deviation of the edge of the sample relative to a reference surface. Typically, this is performed using physical gauges and an overall evaluation of the quality of the strip is made using empirical industry accepted formulae. These measuring techniques, however, are relatively laborious and subjective, and do not lend themselves to continuous monitoring and a quantative evaluation of the edge.

SUMMARY OF THE INVENTION

It is therefore an object of the present invention to provide a method and apparatus to obviate or mitigate the above disadvantages.

In general terms, the present invention provides a method of evaluating the flatness of a laminar strip by projecting onto the surface of a strip a reference line. A camera is positioned above the strip and images the reference line on the strip. The image is processed to identify deviations of the line from a predetermined configuration.

Preferably, the image may be processed to provide a quantative indication of the quality of the deviation for process control purposes.

In one aspect, the present invention provides a method of monitoring the conformity of a surface of a material to a known topography comprising the steps of projecting a beam of coherent radiation on to the surface of the material to provide a line on the surface, obtaining an image of the line, and determining a deviation of the line from a predetermined configuration to compute a degree of conformity.

In another aspect, the present invention provides a system for measuring the conformity of a surface of a material to a known topography. The system comprises a coherent radiation source arranged to direct a beam of coherent radiation on to a surface of the material to provide a line on said surface. An imaging device obtains an image of the line, and a computing device having a processor receives an input fiom the imaging device and processes the image in order to determine a deviation of the line from a predetermined configuration.

BRIEF DESCRIPTION OF THE DRAWINGS

An embodiment of the invention will now be described by way of example only with reference to the appended drawings wherein:

FIG. 1 is a schematic representation of a process control apparatus.

FIG. 2 is an enlarged view of a portion of the apparatus shown in FIG. 1 with a first sample;

FIG. 3 is a view showing an image obtained with the apparatus of FIG. 1;

FIG. 4 is a view similar to FIG. 2 showing schematically the calibration of the apparatus of FIG. 1;

FIG. 5 is a flow chart showing the steps in a height calibration procedure,

FIG. 6 is a flow chart showing the steps in a width calibration procedure;

FIG. 7 is a flow chart showing the steps in performing a measurement of a sample; and

FIG. 8 is a flow chart showing the steps performed during a standard check.

DETAILED DESCRIPTION OF THE INVENTION

Referring therefore to FIG. 1, a sample of steel strip 10 whose flatness is to be assessed is supported beneath a camera 12. The camera 12 is connected to a computer 14 having image processing software 16. The computer 14 is connected to a user interface 18 and an input device 20 to control the image produced on the interface 18. The computer 14 may also connect to auxiliary communications related to process control systems as indicated generally at 22.

A scanner assembly 26 includes a scanning mirror 28 driven by a galvonometer 29 controlled by the computer 14. A coherent radiation source, e.g., a laser device 30, includes a lens to produce a fan shaped optical beam 34 that produces a line on a surface. The laser 30 is positioned such that the beam 34 intercepts the surface of the mirror 28 and is reflected on to the sample 10 at an angle to the optical axis of the camera 12. The point of impingement of the beam 34 on the sample 10 is controlled by the orientation of mirror 28 so it may scan from one edge of the sample to the other. Alternatively, if preferred, the mirror 28 may be held static to provide a line at a fixed location and evaluate the sample 10 based on that location alone.

The assembly 26 may also be arranged to perform both stationary and scanning image processing of the sample 10 by enabling the selective control of the mirror 28. If a stationary assembly is desired, the assembly would comprise a laser device 30 without a rotatable mirror 28. Such an assembly would only impinge the sample 10 at a single location. However, the sample 10 may also move relative to the laser device 30, eliminating the need for a rotatable mirror 28 whilst enabling the assembly 26 to perform a scan of the sample 10.

The image of the line on the surface of the sample 10 is obtained by the camera 12 and is processed by the computer 14. As shown in solid line in FIG. 2, if the sample 10 is flat, the beam 34 should illuminate a straight line on the surface of the sample and appears as a straight line on the user interface 18.

If however there is unevenness in the surface 10, as illustrated by the dotted line in FIG. 2, the image of the line obtained by the camera 12 deviates from a straight line. The image obtained and displayed is displaced depending on the deviation of the surface from planar and a corresponding image produced on the interface 18, as shown in greater detail in FIG. 3.

To obtain a quantative measurement of the deviation, the image obtained by the camera may be analyzed in one of a number of manners.

The normal steel industry standard to measure deviations in the flatness of a strip product is to utilise what are known as I Units. An I Unit is a function of the wavelength of the disturbance and the height of that disturbance from the ideal planar state. A combination of those two measurements is then utilized to compute an I number accepted within the industry.

As may be seen in FIG. 1, the laser beam 34 is scanned onto the surface through mirror 28 that can be stepped in increments across the surface from one edge to the to other. A measurement of the deviation from flat can be obtained at each scan although, for practical purposes both edges and the centreline are all that is required. Because the angle of incidence of the beam varies from one edge to the other, it is first necessary to calibrate the imaging apparatus to obtain a compensation factor as a function of the angle of the beam.

In order to calibrate the apparatus, as shown schematically in FIG. 4, a block 40 is placed upon a surface with a known height and width. The offset from the image of the line as it passes over the block 40 is noted, and successive blocks are placed with differing heights to provide calibration data This is performed at each position of the mirror 29, providing a set of calibration data that can be used for each position of the mirror 29 in order to calculate the height above the surface corresponding to a particular offset of the image, as well as the width at successive mirror positions. Thus, two separate calibration procedures are performed, namely a height calibration and a width calibration.

The height calibration procedure is shown in FIG. 5, and can be visualized by referring to the schematic of FIG. 4. An operator first places the block 40 on the apparatus table 42. Once the system is initialized, the laser line 34 is moved out of the field-of-view of the camera 12 by sufficiently adjusting the mirror 29, and a background image is captured by the camera 12 and stored for use during subsequent imaging. An array of N image buffers is then created, and each of the N buffers contains an image that corresponds to respective ones of N imaging laser positions along the table 42. N is chosen depending on the nature of the application, and may vary based on the particular accuracy of the calibration data which is desired.

The laser line 34 is then placed at its initial position, and an image is obtained of the laser line 34 passing over the block 40, at each laser position, through an entire region of interest. Thus at each position, an image is obtained and the corresponding background image from the buffers is subtracted from it to help eliminate ambient light noise. A smoothing kernel may then be applied as well as an optional subpixel interpolation. Details of the subpixel interpolation will be described in more detail later. The imaging process is repeated N times to obtain an image of the laser line 34 at each of the N laser positions. The angle of incidence of the laser line 34 decreases whilst the laser line 34 passes over the block and the discontinuity of the line deviates further from its normal configuration. Since the height of the block 40 is known, the software 16 can calibrate the scale at each position.

Once the array of images is obtained, a second loop begins to process each image to calculate the scale at the respective position. The region of interest in the image corresponding to the portion of the line comprising the deviation in the laser line 34 is first identified. A sub-loop then begins wherein for each column of pixels in the image, the centroid of the image is identified by calculating the double derivative of the row of pixels. A median filter of the centroid of the laser line 34 is then applied to remove any erroneous spikes, and the centroid derivative is calculated to reveal the pixel shift due to the calibration sample being in the laser's path (i.e. the deviation of the discontinuity from the normal line). Each iteration of the second loop completes by plotting the pixel shift versus position voltage of the galvonometer, indicating the pixel shift (and thus height) versus the laser position.

Once the second loop is complete, a 2^nd order polynomial is fitted to the plotted points and the solution fit is then stored to a file. The height calibration is then complete, and a width calibration can be performed.

The width calibration procedure is shown in FIG. 6, and comprises steps similar to those implemented during the height calibration procedure until the final step of the second loop. During the second loop, the derivative of the centroid reveals the separation between the calibration sample's edges (i.e. the width of the block). The derivative peak separation versus position voltage is thus plotted to complete each iteration of the second loop. Once the second loop is complete, a 1^st order polynomial curve is fit to the plotted points and the solution fit is then stored to a file. The width calibration is then complete. The two calibration procedures are essentially similar, which allows the program to reuse some of the routines that are executed.

It shall be noted that preferably, different sample block orientations are used for the width and height calibration procedures. For instance, 7 mm wide×51 mm high orientation has been found to be suitable for the height calibration and a 51 mm wide×7 mm high orientation has been found to be suitable for the width calibration. The operator would thus reorient the block between successive procedures to place the dominant dimension along the direction of the impinging line.

The width measurement is substantially linear with respect to the distance of the block 40 from the camera, which allows a 1^st order fit. The height measurement changes as the line moves away from the camera as well as due to the height of the sample off the table 42, which is nearly 1^st order. However, a 2^nd order polynomial is preferably used to provide a better fit. It will be appreciated that the choice of polynomial to fit the data is application dependent and may change based on the software or hardware used.

After calibration, measurement is performed as shown in FIG. 7.

Referring to FIG. 7, in order to compute the I number for a sample 10 interrogated by the system, the sample 10 is first placed on the table 42, preferably against a pair of alignment pins 11 (see FIG. 2). The system is intialized and the laser line moved out of the camera's field-of-view to enable the camera 12 to obtain a background image. As indicated above, typically only three measurements are required, namely the two edges and the center of the sample. The laser line 34 is moved towards one of the edges or the center. Usually only the approximate edge location is known, therefore other image processing may be used to accurately locate the edges or center if desired.

An array of N image buffers are created to record the image at each of the N positions of the laser, typically 3, and a first loop is entered to capture and optionally process the images before analysing the images. The first loop comprises incrementing the laser line position, capturing the image, removing the background image, and optionally but preferably applying a smoothing kernel and subpixel interpolation. The loop iterates N times, one for each laser position.

Once the first loop has completed, the region of interest (ROI) of the laser line in each image is found. One ROI typically captures all lines since the lines would be incremented by only a small distance. A second loop then begins which comprises performing a sub-loop for each column of pixels that finds the centroid by calculating the intensity double derivative of the row of pixels, and applying a scaling operation to the pixel position to shown the height in mm. The double derivative finds the location of the line in each column. The scaling has been calculated in the calibration step to compensate for the particular laser position.

A third loop then begins, which, for each image, median filters the laser line centroid, applies smoothing, calculates I, stores the I value in an array, and calculates the standard deviation of the laser line 34 and stores in an array.

The computation of an I Unit is determined utilising the formula, $\frac{L_m-L_s}{L_m}\times {10}^5$
where L_m is the length of the line as measured on the sample. L_s is the length of the ideal line with a perfectly flat surface.

In order to compute the value of L_m, the deviation of the line from the idealized straight line is measured from the image. This is performed by taking the change in offset position between adjacent pixels and summing them across the length of the line (i.e. L_m=√{square root over (Δx₀²+Δy₀²)}+√{square root over (Δx₁²+Δy₁²)}+ . . . where Δx_i and Δy_i are the incremental changes on the x and y axes respectively at position i). The net deviation is an indication of the increase in length between the ideal straight line and the actual line and is used to compute the I Units at each scanned location.

The standard deviation is used to determine when the laser line 34 has moved off the sample 10. A sudden increase in standard deviation indicates that the line is beyond the edge. Once the third loop completes, the I Units are only considered where the corresponding standard deviation is below a predetermined threshold. This ensures that only relevant images are used to calculate the ultimate I Unit for the sample 10.

A median I is calculated fiom the array of I values stored during the third loop and the results are written to a file.

FIG. 8 shows a standard check procedure. A standard check may be performed at desired time intervals to verify the integrity of the system. The standard check may be used to determined whether a recalibration is required. For example, a standard check may be performed every 12 hours, at the beginning of each shift in a manufacturing environment. The operator can obtain an indication of how well the system is operating and can calibrate prior to operating a line. Calibrations and standard checks may be performed as fiequently as required by the particular application, or based on quality standards.

The subpixel interpolation algorithm step shown in FIGS. 5-8 can be applied to the images after smoothing to determine the lateral extent, i.e. thickness of the laser line on the image. Subpixel interpolation allows the resolution to be enhanced using software and reduces the cost and operating requirements for the hardware.

A typical sub pixel interpolation algorithm involves subdividing pixels based on the intensities of neighbouring pixels. Particularly, what is referred to as “Four Ray Supersampling” involves creating a virtual image by covering the image with a finer grid than is actually available, and then using multiple sample points to determine an intensity for each pixel. For example, using M sub-pixels an average of super-sampling points can be determined by the formula $I^{\prime }\left(i,j\right)=\frac{1}{M}I\left(p,q\right),$
where I(p,q) is the intensity at sub-pixel (p,q); s(i,j) shows all sub-pixels (p,q) of (i,j), i.e. the neighbourhood of the pixel; and I′(i,j) is the final intensity (i,j) at the pixel. Optionally, a weighted average and/or post-filtering may also be performed.

Although the present invention has been described in the context of monitoring the flatness of a sheet of steel 10, it will be appreciated that the invention may apply to other materials where the monitoring of flatness is desired, such as copper, aluminium or plastic. Moreover, although the system has been described having separate computing devices, software, and a camera, that a smait camera may be used to provide data processing capabilities in a single device. It will also be appreciated that the functionality of the software may be implemented using hardware as desired.

It will also be appreciated that the above operations are also applicable to stationary imaging where N=1 and a single image is processed, as opposed to a plurality of images obtained by rotating the mirror 28 and thus scanning the laser line 34 over the sample 10.

The above example describes measuring the flatness of a substantially planar surface. It will be appreciated that the above concepts can be applied to the measurement of other topographies, such as the curved upper surface of a tube. In such an application, the impinging laser would follow the contour of the surface, and when directed at an angle thereto, would create an elliptical line in the image. Through a series of calibrations using a curved block, a desired image can be established and deviations in this line would indicate a deviation from the roundness of the surface. Other surface topographies could also be monitored and measured using the same principles by applying a series of calibrations and devising a predetermined datum.

Although the invention has been described with reference to certain specific embodiments, various modifications thereof will be apparent to those skilled in the art without departing fiom the spirit and scope of the invention as outlined in the claims appended hereto.

Claims

1. A method of measuring the conformity of a surface of a material to a known topography comprising the steps of projecting a beam of coherent radiation on to said surface of said material to provide a line on said surface, obtaining an image of said line, and determining a deviation of said line from a predetermined configuration to compute a degree of conformity.

2. A method according to claim 1 wherein said method measures the flatness of said surface, and said degree of conformity is a degree of flatness.

3. A method according to claim 1 wherein said line is scanned along said surface and a plurality of images are obtained at predetermined positions of said line.

4. A method according to claim 3 wherein said beam is directed towards a mirror and reflected by said mirror onto said surface, and wherein rotation of said mirror scans said line along said surface.

5. A method according to claim 2 wherein said material is steel and said degree of flatness is measured in I Units.

6. A method according to claim 1 wherein prior to projecting said beam, a background image of said material is obtained, said-background image being subtracted from said image to remove ambient light noise from said image.

7. A method according to claim 1 further comprising the step of applying a subpixel interpolation on said image prior to determining said deviation to determine the lateral extent of said line.

8. A method according to claim 7 wherein said subpixel interpolation comprises subdividing pixels of said image based on the intensities of neighbouring pixels and determining an average intensity of said pixels based on said neighbouring pixels.

9. A method according to claim 1 wherein deviation of said image is evaluated on the basis of the number of pixels traversed by said line.

10. A method according to claim 9 wherein the number of pixels traversed by said line is measured from a centroid of said line.

11. A method according to claim 10 wherein said centroid is measured by first locating a region of interest of said image containing said centroid.

12. A method according to claim 9 wherein the number of pixels traversed by said line is converted to a unit of distance in one of millimetres and inches, based on a scaling factor determined by a calibration procedure executed prior to projecting said beam.

13. A method according to claim 12 wherein said unit of distance is used to determine said degree of conformity.

14. A method according to claim 13 wherein prior to determining said degree of conformity, said method further comprises a step of processing said image.

15. A method according to claim 14 wherein said processing includes one or more of applying a median filter and smoothing.

16. A method according to claim 3 further comprising the step of calculating the standard deviation of the line on each said image in said scan, wherein each said standard deviation is compared to a threshold, and if said threshold is exceeded, said image is not used in determining said degree of conformity.

17. A method according to claim 16 further comprising the step of calculating a median degree of flatness from a collection of values of the degree of flatness obtained from each image in said scan.

18. A method according to claim 1 further comprising a calibration procedure prior to projecting said beam.

19. A method according to claim 18 wherein said calibration procedure comprises the steps of inserting a block of known dimensions at an intended location of said material; projecting said beam on to said block and an underlying surface along one of said dimensions; obtaining an image of said line; determining a deviation of said image from a predetermined configuration, said deviation altered according to the extend of said one of said dimensions and being evaluated on the basis of the number of pixels traversed by said line; and calculating a scale correlating pixels of said image to a standard unit of distance using said deviation and said one of said dimensions.

20. A method according to claim 19 wherein said calibration procedure is repeated a plurality of times for a plurality of locations of said line as said line is scanned along said block.

21. A method according to claim 19 wherein the number of pixels traversed by said line is measured from a centroid of said line.

22. A method according to claim 21 wherein said centroid is measured by first locating a region of interest of said image containing said centroid.

23. A method according to claim 20 wherein said one dimension is the height of said block.

24. A method according to claim 23 wherein said scale at each location is mapped to a 2nd order curve, said curve indicating said scale at each location.

25. A method according to claim 20 wherein said one dimension is the width of said block.

26. A method according to claim 25 wherein said scale at each location is mapped to a 1st order curve.

27. A system for measuring the conformity of a surface of a material to a known topography, said system comprising:

a coherent radiation source arranged to direct a beam of coherent radiation on to a surface of said material to provide a line on said surface;

an imaging device for obtaining an image of said line; and

a computing device having a processor to receive an input from said imaging device and process said image in order to determine a deviation of said line from a predetermined configuration.

28. A system according to claim 27 wherein said system measures the flatness of said surface.

29. A system according to claim 27 further comprising a table for supporting said material.

30. A system according to claim 27 wherein said material is steel and said computing device uses said deviation to compute a degree of flatness measured in I Units.

31. A system according to claim 27 wherein said imaging device is a camera.

32. A system according to claim 31 wherein said camera is a smart camera and said computing device and said processor are operated by said smart camera.

33. A system according to claim 27 wherein said beam is redirected towards a plurality of locations on said surface enabling said line to scan said material and said imaging device to obtain a plurality of images at predetermined positions of said line.

34. A system according to claim 33 wherein said processor receives a plurality of images from said imaging device, each of said plurality of images being processed to determine said deviation of said line in each image, said deviations used to determine a representative deviation for said material.

35. A system according to claim 33 further comprising a mirror for reflecting said beam onto said surface, said mirror being rotatable about an axis wherein rotation of said mirror scans said line along said surface.

36. A system according to claim 35 further comprising a motor for rotating said mirror, said motor controlled by said computing device.

37. A system according to claim 27 further comprising an interface connected to said computing device enabling an operator to view and interact with said image.

38. A system according to claim 27 wherein said computing device is connected to an auxiliary interface enabling interaction between said system and an auxiliary entity.

39. A system according to claim 38 wherein said auxiliary entity is a process control system.

40. A system according to claim 39 wherein said process control system further comprises an interface enabling a user to view and interact with said image and said computing device.