AU571673B2

AU571673B2 - Optical article measuring system

Info

Publication number: AU571673B2
Application number: AU41143/85A
Authority: AU
Inventors: Hendrik Charles Van Hoek
Original assignee: Broken Hill Pty Co Ltd
Current assignee: OneSteel Manufacturing Pty Ltd
Priority date: 1984-03-09
Filing date: 1985-03-08
Publication date: 1988-04-21
Anticipated expiration: 2005-03-08
Also published as: AU4114385A

Description

OPTICALARTICLEDIMENSIONMEASURINGSYSTEM This invention relates to a method and apparatus for calibrating a sensor means for measuring the dimension of an article, a method and apparatus for measuring the dimension of the article as well as methods and apparatus for aligning a sensor means and detecting the location of an edge in an image of an object. Quality control and yield optimism of manufactured articles such as steel slabs requires the monitoring of, for example, hot slab dimensions. Traditionally, these dimensions are measured by a manual technique using callipers. This technique requires the slabs to be stationary, and mill personnel to come uncomfortably close to the hot material to perform the measurements. It is therefore apparent that manual techniques have many disadvantages. Modern electronic technology has made possible development of non-contact in line systems to measure the dimensions of hot steel products. For example a method of measuring the dimensions of a slab is disclosed in U.S. Patent No. 4,271,477 issued to Gerald B. Williams. In this patent sensors, such as in line cameras, are utilized to detect the slab. The cameras are focused onto the slab so that the images of the slab are formed on diodes within the camera. By determining the number of diodes which are illuminated within the camera with the image of the slab the dimension of the slab can be calculated by suitable processing circuitry.

In order to obtain an accurate measurement with prior art methods it is necessary to ensure that the photodiode array in the camera is parallel to the slab face being measured. If the array is not parallel to the slab face being measured the conventional techniques will give incorrect measurement values for the various dimensions of the slab due to perspective distortions. Perspective distortions are distortions due to non- linearity of relative displacements between object coordinates in the object plane with respect to relative displacements between corresponding image co-ordinates in the image plane. It is extremely difficult to ensure that the cameras are arranged with their photodiode arrays exactly parallel to the slab face being measured and the object of this invention is to provide a calibration method which overcomes the need to ensure strict parallel arrangement of the array relative to the slab face. Furthermore the present inventors have found that in arranging the photodiode array exactly parallel to the slab face a large portion of the viewing range of the camera is wasted. By tilting the array so that it is not parallel to the slab face the camera can be moved closer to the slab so that viewing range is not wasted and the method according to the first aspect of the invention will automatically compensate for any distortions which are inherent when attempting to align the array parallel to the slab face or when arranging the array at an angle to the slab face so that viewing range of a camera is not wasted.

The invention in a first aspect may therefore be said to reside in a method of calibrating a sensor means for use in measuring the dimension of an article by arranging the sensor means so that it will detect the article to be measured during a measurement step, the method of calibrating the sensor comprising locating a reference datum having at least five reference locations so that the at least five reference locations can be detected by the sensor means, at least three of the reference locations falling in a straight line (as defined herein) and at least two of the reference locations falling in a second line parallel to the said straight line, said second line being spaced from the straight line in the direction of an imaginary line between the sensor means and the reference datum; determining calibration values having regard to the image locations of the reference locations in the sensor means utilizing known displacements between reference locations in the datum to take into account perspective distortions due to the positional relationship between the sensor means and the datum.

A straight line is defined herein to be a line which is straight when viewed from all directions. Since, according to the first aspect of the invention, the sensor means is calibrated to compensate for perspective distortions due to the positional relationship between the sensor means and the datum the sensor means need only be arranged so that they view the article to be measured and the determined calibration values may be utilized in the calculation of the desired dimension of the article to provide improved accuracy in the measurement of that dimension. The invention in the first aspect may also be said to reside in a system for calibrating a sensor means for use in measuring the dimension of an article, comprising a reference datum having at least five reference locations all of which can be detected by the sensor means, at least three of the locations falling in a straight line (as defined herein) and at least two of the locations falling in a second line parallel to said straight line, said second line being spaced from the straight line in the direction of an imaginary linebetween the sensor means and the reference datum, said sensor means in use detecting said reference locations, and processing means for determining calibration values having regard to the image locations of the reference locations in the sensor means utilizing the known displacements between reference locations in the datum to take into account perspective distortions due to the positional relationship between the sensor means and the datum.

Preferably the step of determining calibration values comprises the step of forming equations, which include the calibration values, indicative of the displacement of an object coordinate from the projection of a line perpendicular to a sensing array in the sensor means related to its image coordinate, solving those equations to determine the calibration values, storing the calibration values for use in an equation or equations which give the dimension or dimensions of the article taking into account the distortions. Preferably the method of calibrating the sensor means also determines scaling of displacements between image coordinates as compared to displacement between corresponding object coordinates and also determines unknown relative displacements between the sensor means and a further sensor means. The first aspect of the invention may further reside in a method of determining the dimension of an article comprising arranging at least one sensor means to detect the article when the article is in a measurement position, locating a reference datum at the measurement position so that it is detected by the sensor means to calibrate the sensor means by determining calibration equations, including calibration values, for compensating for perspective distortion and displacement of the sensor means from the datum, solving the calibration equations to determine the calibration values, storing the calibration values, removing the reference datum, and then detecting said article with the sensor means to obtain information concerning the article and calculating the dimension or dimensions of the article by means of measurement equation or equations which include the calibration values and the information concerning the article.

Preferably one of the calibration values is indicative of the displacement of a reference location in the datum from the projection of a line perpendicular to a sensor array in the sensor means related to its image coordinate in the sensor array.

A further difficulty in measuring dimensions of an elongate object such as a hot slab is that the hot slab may exit a rolling mill and travel along a roll table at a slight angle to the longitudinal axis of the table. In order to correctly determine the width of the slab it is therefore necessary to determine the skew angle of the slab with respect to a known axis such as the longitudinal axis of the table.

The method described above of calibrating the sensor means to compensate for distortions due to physical relationships, whilst overcoming significant problems experienced in the prior art, will not compensate for the possibility of inaccurate measurement of at least one dimension of a slab due to the slab being skew with respect to the longitudinal axis of a roll table.

The object of a second aspect of the invention is therefore to provide a method and apparatus for accurately measuring the dimension of an elongate article not withstanding the fact that it may be at an angle with respect to a predetermined axis.

The invention in a second aspect, may therefore be said to reside in a method of measuring the dimension of an elongate article comprising the steps of arranging at least two sensor means to detect said article, calibrating each said sensor means to compensate for distortions due to the physical relationship between the sensor means and the article, receiving information from each said sensor means to determine if the elongate article is at an angle to a predetermined axis and utilizing said information to provide said dimension or allow the dimension to be obtained therefrom. The invention in the second aspect may also be said to reside in a system for measuring the dimension of an elongate article comprising at least two sensor means to detect said article, processing means for calibrating each said sensor means to compensate for perspective distortions due to the physical relationship between each sensor means and the article and for receiving information from said sensor means to determine if the elongate article is at an angle to a predetermined axis for use in providing said dimension or allowing the dimension to be obtained therefrom.

Preferably the elongate article is a slab and the predetermined axis is the longitudinal axis of a slab table on which the slab is supported.

Preferably two sensor means are arranged above the plane in which the article skews, said sensor means being displaced relative to one another, along the predetermined axis, by a predetermined distance and the skew angle of the article is determined from the distance between the sensor means and the distance between each sensor means and a central point of the slab along a line perpendicular to the predetermined axis.

The invention also provides a reference datum for use in calibrating sensor means to be used to measure the dimensions of an article, said datum comprising a support frame, said support frame supporting at least five reference locations such that three reference locations are arranged in a straight line which will be transverse to an imaginary line between the reference datum and the sensor means when the reference datum is in use, and two of the reference locations being in a line parallel to said straight line and spaced from said straight line in the direction of said imaginary line.

In a preferred embodiment of the invention will be illustrated in conjunction with the measurement of dimensions of a hot slab produced in a steel mill with reference to the accompanying drawings in which:- Figure 1 shows a schematic view of a system for measuring dimensions of a hot slab;

Figure 2 shows viewing geometry of two of the cameras used in Figure 1; Figure 3 is a plan view of a slab skew measurement arrangement;

Figure 4 is a view of camera viewing geometry and illustrates a comparison between viewing geometry of a parallel array and a tilted array; Figure 5 is a end view of reference datum in the form of a calibration frame on a roll table;

Figure 5A is a side view of the frame of Figure

1;

Figure 6 is a diagram of viewing geometry seen by one of the cameras when viewing the calibration frame of Figure 5 and 5A;

Figure 7 is a view of camera viewing geometry used in calibrating the camera;

Figure 8 is a view of camera viewing geometry for two cameras used in calibrating the cameras;

Figure 9 is a plan view of a skew calibration arrangement;

Figure 10 is a diagram showing a degraded and ideal image response of a slab edge; Figure 11 is a block diagram of a processing system;

Figure 12 is a flow chart of a main control program;

Figure 13 is a view of a camera alignment system; and

Figure 14 is a side view of the system of Figure 13.

With reference to Figure 1 three cameras numbered 1 to 3 are shown arranged around a hot slab. The cameras are interconnected with a microprocessor system 12. Cameras 1 and 2 are arranged to enable the thickness and the width of the hot slab 10 to be determined. Camera number 3 is provided to enable compensation for any skew of the slab 10 on a roll table. The cameras 1 to 3 are preferably line scan cameras which consist of a normal camera lens which focuses an image of an object onto a linear array of photodiodes. The image received by the photodiodes therefore represents only a single line or a narrow band of the object in view. The photodiodes produce an electrical signal proportional to the intensity of incident light. Additional electronics within the camera samples the signal produced by each photodiode serially and produces an electrical (video) signal varying in time. This latter signal is therefore a facsimile of the image intensity varying along the length of the array.

The cameras are all aligned so that their lines of view are across the appropriate slab face and perpendicular to the direction of travel of the slab. The apparent width of a slab face as determined from the location of the slab edge images senses by each of the cameras is dependent on both the true width of the face and its distance from the camera. Therefore to measure the sectional dimensions of a slab it is necessary to determine the distances from each camera to the slab faces. This is achieved by taking simultaneous measurements from the two cameras viewing the top and side faces of the slab in the same plane. The dimensions are obtained by solving the following simultaneous equations which are derived from Figure 2 below. Note that the distances dx and dy are not the nominal focal lengths of the lenses, but are the distance from the principal point of the lens to the array.

Slabs lying skew on the roll table will cause an error in the width measurement of the slab because the viewing line of camera 2 will no longer be parallel to the width dimension. The skew does not affect the thickness dimension of the slab. The relationship between the true width W and the apparent width W' for a skew angle β is:

W = W' cos β (v)

Determination of the amount of skew and hence correction of the apparent slab width is achieved by employing a third camera which views the top face at a known distance from camera 2 as illustrated in Figure 3. The skew angle is defined by the following expression. X_m and Z_m are the distances along the camera viewingg lines between the middle of the top face to camera 2 and 3 respectively. Z_d is the displacement between cameras 2 and 3.

The above equations (i) to (vi) are valid only if the camera arrays are aligned parallel to the slab faces in view. This would require a critical alignment of the cameras which is difficult to achieve in practice. In addition, such an alignment would not necessarily make the best use of the available measuring range of the cameras. For example, if the array of camera 1 was aligned parallel to the side face of the slab and located at a height of 50mm above the bottom of the slab, to avoid viewing the top or bottom faces when locating the edges of the thinnest slab of 100mm, its total viewing range would need to be 500mm in order to view a maximum thickness slab of 300mm. This wasted viewing range and hence loss of resolution is avoided by tilting and moving the camera with the same lens closer to the slab face to achieve a viewing range of 300mm as seen in Figure 4. Any angle between the sensing array and the face in view introduces a perspective error. The effect of perspective is indicated by the dependence of the apparent dimension of an image on its position on the sensing array.

To overcome the problems associated with the need to arrange the sensing arrays of the camera parallel to the slab face in view the invention provides a calibration technique to compensate for the effect of perspective distortions introduced by the physical relationship between the cameras and the slab.

With reference to Figure 5 and Figure 5A calibration frame 20 is provided which has a number of reference points disposed thereon. Reference points 22 to 30 are arranged in a single plane so that they can be viewed by one of the cameras arranged above the calibration frame, for example camera 2 in Figure 1. Reference points 32 to 40 are arranged so that they are viewed by camera 1 in Figure 1. The calibration frame 20 may include legs (not shown) for securely supporting the calibration frame 20 on the roll table so that the calibration frame is aligned perpendicular to the direction of travel of a hot slab. The frame 20 comprises four planes which contain the reference points 20 to 40. Reference points 22 to 26 are arranged in one plane and reference points 28 and 30 are arranged in another plane for viewing by camera number 2 in Figure 1 and reference points 32 and 34 are arranged in one plane and reference points 36 to 40 are arranged in a further plane for viewing by camera number 1 in Figure 1. The first two mentioned planes are parallel to the top face of a slab and the other two planes are parallel to the side face of a slab.

The reference points are preferably in the form of lights which are provided in housings so that they direct light only towards the camera by which they are intended to be viewed. The reference points 22, 24, 26, 28 and 30 should not be viewed by camera number 1 and the other reference points should not be viewed by camera number 2. Reference points preferably have a width of about 3mm and are somewhat longer in length. The image of the reference points therefore appear as very sharp responses which approximate a normal distribution. The centre of the reference point image can therefore be located by finding the mean of the distribution. As the width of the distribution is spread over a number of photodiode elements the reference point centre can be located to an accuracy better than resolution of the camera.

To calibrate cameras 1 and 2 the reference frame is positioned in the viewing plane of these cameras. The cameras are each adjusted to view five illuminated reference points, three in the nearest and two in the furthest planes. This viewing geometry is illustrated in Figure 6.

The distance of an object coordinate y_i from the projection of the line which is perpendicular to the array is related to its image coordinate m_i by the following expression. u

Where x = distance from the principal point of the camera to the plane of measurement (distance dimensions)

A = constant (array coordinate dimensions) dependent on viewing perspective and is zero when the array is parallel to the plane of measurement

B = constant (array coordinate dimensions squared) dependent on the distance from the principal point to the array K = scaling constant (array coordinate dimensions)

m_m = array coordinate corresponding to the line drawn from the principal point of the lens which intersects the array at right angles. This is nominally the midpoint of the array.

The above mentioned equation is derived in the following manner with reference to Figure 6 and 7.

Referring to Figure 7 :

F = principal point of the lens

OFm_m = line intersecting array at right angles HFm_p = line intersecting object plane at right angles m_i = projection of the object coordinate Y_i onto the array

OF = HF/_cosθ = x/cosθ

OB = OA cos^θ

AB = OA sin^θ BY_i = AB tan ( β + θ ) = OA sin^θ tan (β + θ )

OY_i = OB + BY_i = OA cos^θ + OA sin^θ tan (β + θ )

y_i = OY_i = distance to object coordinate Y_i from 0 in the object plane.

Determination of the unknown quantities in equation Al(i) is possible by observation of 5 reference points in Figure 5 located in two parallel object planes whose relative displacements are accurately known.

The known displacements y₁, y₂, and Δx and the array coordinates m₁ to m₅ allow the following three equations to be formed.

Let m_m-m_i=M_i for convenience M

Combining Al(ii) and Al(iii) to eliminate y₁,

where A = M₁ -2M₃ + M₅ Al(v) and B = 2M₁M₅ - M₃M₅ -M₁M₃ Substituting Al(v) into Al(i) gives

Equations Al(iii) and Al(iv) can now be rewritten in terms of the constants A,B and K as follows:

r

Solution of equations Al(vii) and Al(viii) yields the following expressions for K and x.

The expression for θ can be determined from the expression for K given in equation A(vi). However, K has a dependence on the array coordinate m_m which is accurately known. m_m is assumed to be the midpoint of the array for convenience in the above calculation. The equation for the displacement between two points in the object plane is given by an expression of the form of equations Al(vii) and Al(viii). These expressions and equation Al(x) can be shown to be totally independent of m_m. The displacement of an object coordinate Y_j from the point of intersection of a line, drawn from an unknown array coordinate m_p through the principal point, which intersects the object plane at right angles is given by the following expression which can be shown to be independent of m_m.

The value of the expression for P can be determined from knowledge of the relative displacement of 2 reference points located in displaced object planes. From Figure 6:

To complete the calibration of the two cameras 1 and 2 requires a calibration frame arrangement as illustrated in Figure 8. The known displacements Y₁, Y₂, X_D and Y₁₄ are substituted respectively for the terms y₁, Y₂, Δx and y₁₄ in preceding equations to determine the calibration constants A_y, B_y, K_y, X_x and P_y for camera 1. Similarily the known displacements X₁, X₂, Y_D and X₁₄ are substituted for the terms y₁, y₂, B_x and y₁₄ in the preceding equations to determine the calibration constants A_x, B_x, K_x, X_y and P_x for camera 2. In addition to these known displacements, the displacements of the points in the planes facing one camera with respect to the planes at right angles X_c and Y_c are also known. A prerequisite for the measuring equations in the following section is the determination of the relative displacements between the two cameras X_T and Y_T which are represented by equations Al(xiii) and Al(xiv) below. The subscripts x and y denote horizontal and vertical parameters respectively.

The equations for the slab section dimensions which are derived from equation Al(vii) in terms of the array coordinates and displacements illustrated in Figure 2 are shown below.

y

The terms X₁ and Y₁ which are derived from equation Al(xi) are given by the following expressions.

Solution of equations Al(xvii) and Al(xviii) yield the following expressions which are substituted into equations Al(xv) and Al(xvi) to give the slab dimensions. ^v

To calibrate the third camera, the same calibration frame 20 is transposed from the first calibration position a distance Z_D corresponding to the nominal longitudinal displacement between cameras 2 and 3. The frame is positioned such that its plane remains perpendicular to the direction of travel of the slab and there is no lateral displacement from its previous position. Camera 3 is adjusted to view the five illuminated reference points in the two horizontal planes.

The calibration constants for camera 3 (A_z, B_z,

K_z, P_z and Z_T) are determined in a similar manner to those for cameras 1 and 2. In addition to these constants it is necessary to determine the amount of lateral offset X_o between cameras 2 and 3 as shown in

Figure 9. X_o is determined by finding the difference in the lateral displacements of the two cameras with respect to camera 1. The equation for the offset X_o becomes:

Where X_c = displacement of the first reference point from the side of the calibration frame

X_x displacement of camera 1 from the side of the calibration frame

Z_z = displacement of camera 3 from the top of the calibration frame

Equation (vi) for the skew angle tangent therefore becomes:

Since in the preferred embodiment of the invention the article being measured is a hot slab it is not necessary to provide additional illumination. However if a cold article is being measured additional illumination such as conventional front or back lighting or the like could be provided. Furthermore since the temperature of hot slabs can vary between 900°C and 1200ºC, the illumination received by the camera can change by a factor of more than 20. The preferred embodiment of the invention may therefore compensate for this by adjusting the camera exposure time. The microprocessing circuitry 12 may adjust the clocking frequency of the cameras to maximize the video signal obtained without saturation.

The sharpness of the edges of the slab images obtained by the cameras is degraded by two factors. The first of these is a defocussing effect which results from changes in the object distance from the camera causing shifting of the focused image plane. The range of distance for which a lens maintains a suitably sharp image is commonly referred to as its depth of field and is dependent on the aperture setting.

The second factor influencing image edge sharpness is caused by the radiation emitted from the slab not being constant for the full width of a slab face. The temperature of the slab reduces significantly towards the edges of the face resulting in a gradual reduction in intensity of the corresponding image.

The above effects can be summarized by Figure 10 where the ideal image of the slab edge would be a step. However, as the higher frequency spatial components are removed by defocussing or the overall amplitude of the edge region is reduced by reduction in temperature the location of the steepest gradient in the edge region is unchanged. Therefore the preferred form of the invention incorporates multi-level digitization of the analogue video signals from the cameras and subsequent software processing by the microprocessor to determine the location of steepest slope in the edge region of the image. The main processing unit 12 shown in Figure 11 incorporates an Intel 8085 microprocessor with associated program memory, camera interface, analogue-to-digital convertors (ADC's), direct memory access (DMA) type memories, serial interfaces, a front panel keyboard and display, and power supplies. The camera video signal is digitized by the ADC and stored in memory via DMA. The microprocessor processes this stored video data to calculate the required dimensions which are displayed on a front panel. This information is transmitted to a remote display and a computer or other logging device via a RS232C standard serial communication link. The camera interface includes camera clock control logic and digital signal receivers which are used to control the timing of digitization of the camera video signals. Operator intervention of the processing system is achieved via a front panel keyboard whereby other functions such as calibration can be initiated. The calibration constants are stored in non-volatile memory. Additional front panel indication is provided to display any power supply failure or camera over temperature alarms which are also monitored by the microprocessor.

The preferred embodiment of the invention may also include hot metal detectors which are used by the system to establish the position and direction of travel of the hot slab on the roller table. For example two hot metal detectors can be positioned on either side of the camera 1 and 2 viewing plane so that the system will only measure the section size of the slab excluding ends which have not been scarfed. The system software is approximately 6K bytes in length and is written in Intel 8085 Assembler Language. It is structured into a number of modules which are exercised as required by a main control programme which is automatically entered into after the system is powered up. This programme, a flowchart of which is shown in Figure 12 initiates the collection of digitized video data from the cameras and processes this information to calculate the slab dimensions. The system software also contains other functions which can be initiated from the front panel keyboard. These functions include:

- examine and modify system memory contents - examine and modify system input-output ports - restart measuring (i.e. main control programme) - system hardware test - calibrate cameras 1 and 2 - calibrate camera 3 - collect digitized video data - copy video data into non-volatile memory

Another programme which controls the operation of the front panel or console keyboard and display also permits a dynamic display of system memory contents to assist in monitoring the system during trouble shooting. When this facility is invoked the system continuously updates the console display with the contents of a selected memory location while the system is measuring. The selected memory location can also be changed during operation via the keyboard.

A serial communication link from the system to a host computer has been provided to allow collection and monitoring of sizing data. The system automatically sends messages at the beginning and end of each slab, and continuously sends new measurement results when available. The also responds to control instructions from the computer to send a status message, the last measurement result or to reset after an error. The preferred embodiment also proposes a method of aligning the cameras 1, 2 and 3 so that they view the reference points 22 to 40 on the calibration frame 20. As shown in Figures 13 to 14 each of the cameras 1 to 3 (for example camera number 1) is provided with a laser 50. The laser 50 is arranged and fixed relative to the camera so that the beam of light which spreads out in one dimension from the laser 50 intersects with a target, such as the reference points on frame 20, corresponds to the cameras field of view. The laser beam, therefore gives a visual indication of where the line scan camera is aimed. The beam may be spread by the use of electro or acousto-optic scanners or a cylindrical lens, the last being the simplest to implement.

In practice the laser 50 and camera 1 are aligned prior to installation. To preserve this alignment, brackets (not shown) holding them are locked in position. The laser/camera assembly can then be adjusted until the laser light falls on the desired portion of a target. The camera will then be automatically positioned correctly.

To initially align the laser 50 and the camera 1 the camera 1 is leveled and aimed at a suitable test target 52 as shown in Figure 14. The target 52 is set to the same height as the camera so that the field of view is centralized. This is accomplished by monitoring the video output of the camera. The target 52 is moved up and down until it exits the cameras field of view. The target is then positioned halfway between these exit points. The laser 50 is turned on and is adjusted so its beam lies on the centre of the target. The laser camera assembly is moved vertically so that the target moves in and out of the cameras field of view. By monitoring the video output and the laser line, a check can be made of the coincidence of the target moving beyond both the laser line and the cameras field of view. The assembly is repositioned at a lesser distance from the target and coincidence is again checked. Fine tuning of the laser alignment may be necessary. This step is repeated for other distances until coincidence occurs over the entire operation distance.

This technique is suitable for any application which uses a line scan camera without a view finder or one with a view finder where its use is complicated by mechanical constraints or poor locations. This aspect of the invention therefore provides a simple and effective manner of aligning the cameras, for example, to view the reference points on the calibration frame 20.

Since modifications within the spirit and scope of the invention may readily be effected by persons skilled within the art, it is to be understood that this application is not limited to the particular embodiment described by way of example hereinabove.

Claims

THE CLAIMS DEFINING THE INVENTION ARE AS FOLLOWS:- 1. A method of calibrating a sensor means for use in measuring the dimension of an article by arranging the sensor means so that it will detect the article to be measured during a measurement step, the method of calibrating the sensor comprising locating a reference datum having at least five reference locations so that the at least five reference locations can be detected by the sensor means, at least three of the reference locations falling in a straight line (as defined herein) and at least two of the reference locations falling in a second line parallel to the said straight line, said second line being spaced from the straight line in the direction of an imaginary line between the sensor means and the reference datum; determining calibration values having regard to the image locations of the reference locations in the sensor means utilizing known displacements between reference locations in the datum to take into account perspective distortions due to the positional relationship between the sensor means and the datum.
2. A system for calibrating a sensor means for use in measuring the dimension of an article comprising a reference datum having at least five reference locations all of which can be detected by the sensor means, at least three of the locations falling in a straight line (as defined herein) and at least two of the reference locations falling in a line parallel to the straight line, said second line being spaced from the straight line in the direction of an imaginary line between the sensor means and the reference datum, said sensor means in use detecting said reference locations, and processing means for determining calibration values having regard to the image locations of the reference locations in the sensor means utilizing the known displacements between reference locations in the datum to take into account perspective distortions due to the positional relationship between the sensor means and the datum.
3. The method according to Claim 1 wherein the step of determining calibration values comprises the step of forming equations, which include the calibration values, indicative of the displacement of an object coordinate from the projection of a line perpendicular to a sensing array in the sensor means related to its image coordinate, solving those equations to determine the calibration values, storing the calibration values for use in an equation or equations which give the dimension or dimensions of the article taking into account the distortions.
4. The mekthod of Claim 1 wherein the method of calibrating the sensor means also determines scaling of displacements between image coordinates as compared to displacement between corresponding object coordinates and also determines unknown relative displacements between the sensor means and a further sensor means.
5. A method of determining the dimension of an article comprising arranging at least one sensor means to detect the article when the article is in a measurement position, locating a reference datum at the measurement position so that it is detected by the sensor means to calibrate the sensor means by determining calibration equations, including calibration values, for compensating for perspective distortion and displacement of the sensor means from the datum, solving the calibration equations to determine the calibration values, storing the calibration values, removing the reference datum, and then detecting said article with the sensor means to obtain information concerning the article and calculating the dimension or dimensions of the article by means of measurement equation or equations which include the calibration values and the information concerning the article.
6. A method of measuring the dimension of an elongate article comprising the steps of arranging at least two sensor means to detect said article, calibrating each said sensor means to compensate for distortions due to the physical relationship between the sensor means and the article, receiving information from each said sensor means to determine if the elongate article is at an angle to a predetermined axis and utilizing said information to provide said dimension or allow the dimension to be obtained therefrom.
7. A system for measuring the dimension of an elongate article comprising at least two sensor means to detect said article, processing means for calibrating each said sensor means to compensate for perspective distortions due to the physical relationship between each sensor means and the article and for receiving information from said sensor means to determine if the elongate article is at an angle to a predetermined axis for use in providing said dimension or allowing the dimension to be obtained therefrom.
8. The method of Claim 5 wherein the elongate article is a slab and the predetermined axis is the longitudinal axis of a slab table on which the slab is supported.
9. The method of Claim 5 wherein two sensor means are arranged above the plane in which the article skews, said sensor means being displaced relative to one another, along the predetermined axis, by a predetermined distance and the skew angle of the article is determined from the distance .between the sensor means and the distance between each sensor means and a central point of the slab along a line perpendicular to the predetermined axis.
10. The method of Claim 4 wherein at least one calibration value is indicative of the displacement of a reference location in the datum from the projection of a line perpendicular to a sensor array in the sensor means related to its image co-ordinate in the sensor array.

11. The method of Claim 4 wherein two sensor means are employed and the measurement equations are for measuring thickness and width of the article and are:

0

> where K, A, B, are the calibration values and are constant, the subscripts x and y denoting parameters in two planes which are orthogonel to one and other, the x plane being generally parallel to the thickness dimension and the y plane being generally parallel to the width dimension, and wherein X_T and Y_T are determined from the following equations: where X_c, Y_c, Y_y and X_x are all known displacements,

M₁ and M₂ are the displacements of images of extremeties of the article in the sensor means and P is a calibration constant where the calibration constants A, B, K and P are defined for the following simultaneous equations where: ; *

\

where M₁, M₂, M₄, and M₅ are the displacement of images of the reference points in the sensor means and Y₁, Y₂, Δx, and Y₁₄ are known displacements in the reference datum.
11. A reference datum for use in calibrating sensor means to be used to measure the dimensions of an article, said datum comprising a support frame, said support frame supporting at least five reference locations such that three reference locations are arranged in a straight line which will be transverse to an imaginary line between the reference datum and the sensor means when the reference datum is in use, and two of the reference locations being in a line parallel to said straight line and spaced from said straight line in the direction of said imaginary line.
12. The reference datum of Claim 11 wherein the support frame has three reference locations located in a second straight line parallel to said straight line and two reference locations in a further line parallel to said second straight line and spaced from said second straight line in the direction of a line between the reference datum and a second sensor means which detects the reference locations in the second straight line and the further line.