GB2358979A - Exposure control system for a moving target - Google Patents

Abstract

An exposure control system for a moving target (11) uses measurements of the target speed to determine a gain factor constant by which the video signal from camera (12) is multiplied. This enables the exposure of the scan to be balanced. The gain factor constant is determined as a factor of the maximum speed n , and the actual line speed <I>s</I> of the target. The gain factor constant <I>n</I> is defined as:</MAT> <EMI ID=2.1 HE=11 WI=21 LX=450 LY=962 TI=MF> <PC>where<BR> ```<I>d</I> = 2<I><SP>e</SP></I><BR> and<BR> ```<I>e</I> = <I>INT</I>(log<SB>2</SB>( n /<I>s</I>))

Description

2358979 - 1 EXPOSURE CONTROL SYSTEM The invention relates to a method of

exposure control of an imaging sensor applied to a moving surface.

The reflectivity and variations in reflectivity of a material's surface can be used to determine certain properties of the material. For example, the reflectivity and changes in reflectivity of a sheet of rolled metal or paper can be used to detect defects in the sheet. Reflectivity and changes in reflectivity are measured using an imaging system in which light is directed onto the moving surface, and an imaging system takes pictures of the illuminated surface- The reflectivity of the surface affects the amount of light incident upon the imaging sensor, and so one or more exposure control schemes need to be implemented by the imaging system to correctly expose the sensor.

When the subject being imaged is a moving strip or web, a linescan sensor imaging a single line the width of the web forms a useful basis of an imaging system. The system builds up a two dimensional image by appending successive line images together taken as the web moves under the sensor. It is desirable for the imaging system to produce a line image at a constant down web resolution. To accomplish this the device needs to output line images at a rate determined, amongst other things, by the line speed.

In known exposure control systems the sensor scanning rate is varied in proportion to variations in line speed. The exposure of the sensor is then controlled by one or a combination of the following methods; varying the intensity of the light source illuminating the target; adjusting a mechanical iris; or using electronic exposure control within the sensor.

In another system the sensor is run at a constant, full speed. This means that when the target is running slowly a number of overlapping images of the surface are being taken by successive scans. When new scan data is required the last complete scan is used. The intervening scans represent excess, unrequired data.

Under certain circumstances this method as well as methods using electronic exposure control, can lead to small defects being completely missed.

One important feature of such scanning systems is their dynamic range, that is range of speeds of the moving surface over which they give acceptable results. Large ranges or rapid variations in line speeds traditionally require the light source or iris to have a matching dynamic range and speed. Many applications exceed the limits of light sources and irises.

A further complication is that variations in the reflectivity of the surface as well a variations in line speed both need the sensor's exposure control to be varied to keep the sensor operating around it's optimum exposure level.

r r The present invention aims to address these problems and provide an improved method and system for exposure control of a sensor imaging a moving surface over a large rapidly varying speed range.

The present invention provides an apparatus f or and a method of exposure control for a linescan sensor. The method comprises the steps of using the maximum speed v of the surface, over-sampling the surface at a sampling rate dependent on the actual speeds of the surface, s:v, the sampling rate being less than or equal to that required to monitor the surface moving at its maximum speed v and greater than that required when the surface is moving at half its maximum speed, applying a multiplication factor and averaging'the over-sampled scans.

variations in exposure due to line speed are taken into account and thus they are separated out from the exposure control required to compensate for the reflectivity of the obj ect being imaged. Variations in the exposure of the resultant image can be attributed to the lighting level, the iris position and reflectivity of the subject being imaged. Thus the traditional exposure control methods need only be used to cope with long term variations in the light source and variations in strip reflectivity. Indeed a measure of strip reflectivity is now easily realised from the iris position and/or the lighting level required to correctly expose the camera.

Electronic exposure control is not required and all scans are processed, thus 100% coverage of the web can be 1 guaranteed.

The invention requires no moving parts and is completely predictable over a large dynamic range. Thus it is very easy to accurately track rapid accelerations of line speed and operate over large dynamic ranges.

The camera is always scanned at a rate somewhere between that required to match the maximum line speed and just over half that speed. Electronic exposure control is not used, and so the exposure period, that is the length of time the sensor is exposed, in each sensor scan may vary, depending on the line speed. To compensate for the variable exposure period, each pixel value is multiplied by a suitable value n between.one and two to balance all scans.

For example, if the line speed is exactly half the maximum line speed, then the sensor is set at full speed and two scans are added together. Before this summation, the video has the maximum gain correction of two applied (minimum exposure period, so needs maximum correction). After the summation, the sum is divided by two (because two scans were added together). Thus, in this example, one video scan is output for two input sensor scans.

more generally, if v is the maximum line speed and s is the current line speed, then we can define a multiplication constant n and divide value d (derived from the exponent e) as follows:

T e = INT (1092 (V1 5)) d = 2 e n = 2 dslv The incoming video is multiplied by the constant n. Then, d scans are added together and the accumulated result is divided by d bef ore being output. The result is a video output at a rate that is proportional to surface/web speed and at a level that is invariant of surface/web speed. The camera is always operating within its optimum range while the system provides a large dynamic range.

The sensor is not necessarily correctly exposed. The exposure level is still affected by the material reflectivity, the iris setting and the amount of light incident upon the surface/web. The exposure setting must still be dealt with by another means (e.g. by adjusting the iris or light source level). However, this setting is far easier to accomplish as it is no longer affected by variations in line speed.

A preferred embodiment of the invention will now be described with reference to the drawings in which:

Figure 1 shows a block diagram of the components of a surface inspection system embodying the invention; Figure 2 shows the components of a CCD line scan camera - 6 controller according to an embodiment of the invention; Figure 3 shows the components of an exposure controller according to an embodiment of the invention; Figure 4 shows the operational program of the line speed feedback system in the preferred embodiment of the invention; and Figure 5 shows the operational program of the exposure controller of Figure 2.

In the preferred embodiment, a system for imaging a web 11 being rolled, such as a metal strip, is provided.

The apparatus comprises means for reading the actual speeds of the target surface, means for scanning the target surface at an over-sampled and variable rate, means for multiplying the video by a factor and means for averaging the oversampled video scans.

The target surface is scanned by illuminating it and detecting a reflected light pattern from the surface. This is done using a linescan CCD video camera. No electronic exposure control of the linescan CCD device is used, and so the scan rate and the exposure period are the same. In the present invention, even if the web speed is much less than half its maximum speed, the camera is operated at a rate between that required to match the maximum surface speed v and just over half of that speed. If the web speed is less 1 7 - than or equal to half the maximum speed, then the camera is still set at a rate between that required to match the maximum line speed and just over half that speed, but two or more scans will be averaged. Depending on the speed of the web, for example, if a down web resolution of 1.6mm pixels is required, the camera system must output a scan representing a 1. 6mm strip of the web. If the maximum speed of the web is 14m-s-1, the maximum required camera scanning 14ms-1 rate is 0.0016m = 8750 scans per second.

At this scanning rate each scan has an exposure period of 0.0016 14 1141As. If the web is moving at just over half its maximum speed, say 7.1 ms-1, the camera must slow its scan 7.1 rate down to - 0.0016 = 4437.5 scans per second, almost half that of when the web is moving at its maximum speed. This 0.0016 means each scan has an exposure period of - = 225ps.

7.1 To adjust for the difference in exposure period each scan is multiplied by a multiplication factor, described later, to even out the brightness of each scan. If the web is running at 4ms-1, the camera oversamples 2 INT(log2 ( 14 2 INT (1. 8) =21 = 2 -T- and averages two successive scans to produce a single output scan. The camera rate is adjusted to the rate required when 4x2 the web is moving at Y1 4 ( -T4--) of the maximum line speed, 1 - 8 8 namely 5000 scans per second.

0.0016 The gain factor used to scale the incoming video would then 2x2x4 be set to 14 1.14.

The surface is over-sampled by a power of two (e.g. 1, 2, 4, 8, 16, 32, 64 or 128) The period of each sensor scan depends on the actual speed of the surface. The camera is set so that when the speed of the surface is greater than half the maximum line speed, the camera scans a new strip of the surface on each scan, successive scans covering adjacent strips of the surface, there being no overlap of the surface covered by adjacent scans and no strips of surface left unscanned between successive scans. So long as the surface moves at half or less than half the maximum line speed v a number of successive scans can be averaged to take into account the overlapping nature of successive scans. The camera thus scans the surface at an oversampled rate. To compensate for the variable exposure period (but limited to a factor of two by the oversampling technique implemented), due to changes in camera scanning rate to account for changes in line speed, each pixel of each of the over sampled video scans is multiplied by a value of between one and two, depending on the ratio of the actual line speed s to the maximum line speed v and on the camera speed, to balance out each scan.

1 1.

- 9 A plurality of scans is then averaged to provide an average value for each pixel.

The maximum speed of the moving web 11 is an installation specific variable and is stored in memory in a host computer 20. The means f or reading the actual speed of the web comprises a mill tachometer (hereafter called a tacho) (not shown) which is connected to the system via a tacho interface 60. The mill tacho detects the rotational speed of one more of the rollers supporting the web.

The video camera used is a line-scan Charge Coupled Device (CCD) camera. Line-scan camera takes pictures of a line of pixels, rather than of an area of the surface, and, because the surface moves, successive pictures are of successive 1 strips of the surface.

The required down web resolution may be set at the beginning of the scanning process or may be an installation specific variable, and is stored in memory in the host PC 20. From the required down web resolution and the maximum line speed the minimum sync period can be set The system processes video from the line-scan CCD cameras using the host computer 20 which initialises and controls dedicated hardware. The hardware consists of a set of registers 22, 24, 26 which the host 20 uses to dynamically configure the system in response to line speed changes, a memory block 28 to implement the variable gain (by means of a Look-up table), a column adder 30 and row storage element 32 to sum d successive scans, a barrel shifter 34 to divide 1 - 10 by d, and timing control logic 36.

The registers are a SyncPeriod register 22, a Gain register 24 and an Exponent register 26, the calculation of which is described later in relation to Figure 5. The row storage element 32 is a First In First Out (FIFO) memory device, and the memory block 28 is a 128 Kbyte by 8 bit Dual Port Random Access Memory (DPRAM).

The timing control logic 36 ensures that only complete sets of register data are used together, and aligns gain constants with the appropriate camera scans, controls the timing of row additions and indicates that the output data is valid (output line 56).

The ten bit output of camera 12 is used to address the least significant bits of the address bus of the DPRAM 28.

This memory is used as a Look-Up Table (LUT) to multiply the video levels by the multiplier constant n. The seven most significant address bits of this DPRAM 28 are the Gain input value that sets the gain correction to apply to the video signal. The eight bit data output 58 of this LUT represents the Gain corrected video level. The contents of the LUT are controlled by the host processor 20 via the second port of the DPR.AM 28.

The Gain corrected video 58 is fed into the column adder 30, which, in combination with the FIFO 32, constitutes a 15 bit pixel column summation circuit. This sums up to 128 successive scans after which the sum is reset to zero. The output 50 of the adder 30 is fed into a shifter 34 that shifts right by 0 through 7 bits (divides the result by 1, 2, 4, 8, 16, 32, 64 or 128). The shift is set by the exponent input 52. The output 54 of this shifter 34 is only valid when indicated by the timing control circuit 36 on output 56.

Figure 4 shows the operational program for tracking line speed variations. First the current line speed is read 70 from the tacho interface 60. Then the minimum and current sync periods are calculated 72. The minimum sync period is the down web camera resolution divided by maximum line speed. The current sync period is the down web camera resolution divided by the current line speed. The maximum line speed and down web camera resolution are installation specific variable which are stored in memory by the host PC 20.

The "Sync Ratio" is then calculated 74. This is the log ratio of the required sync period relative to the minimum sync period.

The whole part of the Sync Ratio is then extracted and written 76 to the Exponent Register 26. The fractional point of the Sync register is also extracted and an integer equivalent of this is written 78 to the Gain Register 24. An integer equivalent of the sync period is calculated and then divided by the number of input camera scans the hardware will add together 80. This value is then written to a Sync Period Register 22.

Figure 5 shows the operational program used to initialise the Gain Look-up Table 28. First, the ratio between successive Gain values used by the hardware is calculated, and a loop counter for counting through all integer representations of the Gain is initialised 82. A loop is then entered, at the beginning of which a check is made to see whether the end of the loop has yet been reached 84. If the end of the loop has not yet been reached, the Sync Ratio is calculated 86. This represents one of the possible Sync Ratios that may be found in 74 and will be a number greater than or equal to 1 and less than 2.

The gain to be applied to the input video value is then calculated 88. This takes into account the 8 bit output range (256), the 10 bit input range (1024), the change in exposure relative to the reference sync period (Sync Ratio) and a further factor (2). This last factor ensures that for all sync periods, the full range of output values is used.

The system then sets the Input to zero 90 in preparation for entering another loop. Inside the loop, a check is made to see whether the end of that loop has yet been reached 92. If not, calculations are carried out for each input video level in turn. The output video level for each given input is calculated by multiplying it by the gain. The system then checks that the output video level is within video limits 94. The output level is then written to the location in the Look-Up Table 28 that will be addressed by the given input video level (Input) at the given Gain setting (Loop Counter). The loop is continued until all video levels have been calculated. The Loop Counter is then increased by one.

The system in which the invention has been incorporated utilises a fixed high intensity light source and a mechanical iris to set the basic exposure level.

With the advent of ever faster microprocessors, the system could be implemented entirely within software given a suitable computing platform.

The invention thus provides an exposure control system and method which is predictable over a large dynamic range, and in which exposure control is greatly simplified as line speed variations are automatically taken into account, and the camera is always working within its preferred range of scanning rates.

141 Note to accompany Figure 4 1. All hardware registers are integer values. Registers representing fractional real numbers have an integer written to them that is a multiple of the fractional real number. The multiple is fixed for a given register, but varies from one register to another.

2. For clarity, the flow chart does not include limit and other error checking.

3. The hardware synchronises all register writes to ensure that only a complete matching set of register settings are used together.

1

Claims (8)

1. A method f or exposure control of a scanning imaging sensor aimed at a moving target, comprising the steps of:

a) reading the maximum speed (v) of the moving target; b) detecting the speed (s) of the moving target; c) detecting light reflected from the moving target; d) repeating step c) at a rate between that required to match the maximum speed v of the target and half that speed, (v/ 2); e) converting the light detected into data signals representing the light intensity patterns detected; and f) multiplying the video signal for each scan by a multiplication constant determined, in dependence on the maximum speed (v) and the actual speed (s) of the moving target, to balance the exposure of all the scan.

2. A method according to claim 1 comprising the further step, of averaging the data signals from one or more scans after step (f).

3. A method according to claim 1 or 2 in which data signals are multiplied by a multiplication constant n, defined as follows:

n=2ds/v, where d=21 and e=INT(1092 (VIS)

4. A method according to claim 2 or 3 in which 2e scan are averaged, where e=INT(1092(VIS)).

5. An apparatus for exposure control of an imaging sensor aimed at a moving target, comprising: means for reading the maximum speed (v) of the target; means for detecting the speed (s) of the target; means for detecting light reflected from the target the detection being repeated at a rate between that required to match the maximum speed v of the target and half that speed, (v/2); means for converting the detected light into data signals representing the light intensity patterns detected; and means for multiplying the data signal for each scan by a multiplication constant determined in dependence on the maximum speed (v) and the constant speed (s), to balance the exposure of all scans.

6. An apparatus according to claim 3 further comprising means for averaging the data signals for a plurality of scans.

7. An apparatus according to claim 3 or 4 in which the means for detecting reflected light comprises a video camera.

8. An apparatus according to claim 7 in which the camera is a linescan camera.