GB2248686A - High resolution camera sensor having a linear pixel array - Google Patents

Abstract

The sensor includes a number of sensing elements (e.g. CCD 34) arranged to view a linear array of pixels transverse to the direction of travel of objects to be sensed. Each element is sampled in turn to produce a serial analog output signal, which is supplied to an A/D converter 48. A clocking circuit 35 is synchronised with scanning of the pixel array and transitions in the A/D converter output representative of sensed object edge points are used to gate count values from the clocking circuit for output e.g. to a buffer memory 52 and microprocessor 54. Analog signals from pixels where an object is present may be supplied via a circuit 40 to the A/D converter for sefting the threshold of the analog signal at which a transition in the digital signal occurs. The sensing elements may be CCD's whose conditioning voltage is applied in synchronism with resetting of the clocking circuit and whose outputs are sampled in succession between conditioning voltage applications so as to transfer each pixel charge fully and so as not to allow any charge to accumulate between CCD resets and the next scan. <IMAGE>

Description

HIGH RESOLUTION CAMERA SENSOR HAVING A LINEAR PIXEL ARRAY This invention relates to a camera sensor having a linear array of CCD units that can be used in connection with the eal-time creation of a high resolution silhouette image of an Dbject on a moving conveyor.

Background Information In the inspection by video equipment of a stream of like objects being sequentially transported on a conveyor, it is required that the image processing be done on a real time oasis to produce the necessary secondary control signals.

various prior art techniques are disclosed in Ohyama U.S.

Patent No. 4,866,783 and Turcheck et al U.S. Patent No.

C,784,493.

Composite video signals are not required for some lpplications. Therefore a raster scan is not essential. It nay be sufficient to have a high resolution silhouette of an object elevation to determine e.g. the object orientation or size. Real tine processing of large amounts of data is prohibitive for a feasible low-cost system due to the processing time involved and huge memory requirements to store 1ll the information customarily used. Usual solutions would De to invest in an expensive, faster computer and to add on the required memory.

Summary of Invention The present invention provides a method of scanning an nudge of an object comprising the steps of moving the object elative to a camera sensor comprising a number of sensing elements arranged to view a linear array of pixels extending transverse to the direction of relative movement and spanning the object edge and sampling each sensing element in ;uccession so as to scan the pixel array.

The invention also provides apparatus for carrying out -he above method.

The sensing elements may produce output signals in the orm of a serial data stream of known scan length and the method may further comprise the step of converting the serial data stream into digital signal pulses having leading or railing edges related to a point on the object edge.

The serial data stream may have an analog waveform, which Ls supplied to an A/D converter whose output is analysed to produce said signal pulses at transition points in the A/D inverter output signal. Scanning of the pixel array may be ;ynchronised with a clocking circuit whose output is gated by -he signal pulses, thereby to provide count values related to he position of the object edge point within each scan.

The sensing elements may be light sensitive, such as CD's, all arranged to face a light source with the object dyeing moved relative to the sensor along a path between the Light source and the elements to thereby shade a first portion ?f the pixels, while a second position of the pixels remains sully illuminated. Positions of adjacent pixels where a transition between a shaded and an illuminated pixel occurs ire detected.

Where CCD's are used, a voltage may be applied nomentarily to each CCD unit so as to scan the pixel array; -he output of each CCD unit being sampled in succession between successive said voltage applications so as to transfer sach pixel charge fully and so as not to allow any charge to accumulate between CCD resets and the next scan. The clocking circuit is advantageously reset following each application of -he momentary or conditioning voltage.

Thus the invention may provide a camera sensor for detecting a linear distance in one direction between marginal sedges of an object as the object advances relative to the amera sensor position in a second direction that is transverse to the first direction comprising: at least 1000 sensing elements providing pixels aligned in a single column in a direction parallel to said one lirection; means periodically applying a momentary voltage in timed relation to resetting a counter means; means periodically scanning said pixels to provide an output analog information signal containing variations orresponding to a position of two object marginal edge points; and means responsive only to such signal variations for supplying transfer signals to said counting means to provide a stunt value corresponding to the linear distance between said two edge points during each scanning period.

In a further aspect the invention provides a camera sensor for determining the position of an object edge relative to a known camera position comprising: a plurality of sensing elements providing pixels having density of at least 1000 per inch (394 per cm) aligned in a single linear column which extends across expected location Eor the object edge and which faces a collimated light source; nd means for identifying which of said pixels is aligned vith a point on the object edge which includes: means for scanning said pixels in timed relation with operation of a counter means; an analog to digital conversion circuit connected to receive an analog information signal from said scanning means lnd to provide an output transition signal corresponding to the object edge point location; and means for transmitting said transition signal to record a value in said counter means that identifies a point along said object edge.

Where the object is on a horizontal conveyor, the linear array of pixels may be disposed vertically to provide information conforming to a trace of the upper marginal edge lnd other surface features that appear as a part of a silhouette of the object.

Further features of the invention are in the claims and in the following description made by way of example and with reference to the drawings, wherein: Fig. 1 is a diagrammatic view of a conveyor system for separating and orienting parts, together with a novel inspection camera and information processor; Fig. 2 is a block diagram of a camera sensor and related functional circuitry for acquiring and storing object silhouette information; Fig. 3 is an elevation of a conveyor moving surface that is supporting a round of ammunition; Fig. 4 is a group of waveforms taken at scan position 120 as depicted by line 4-4 of Fig. 3; Fig. 5 is a group of waveforms taken at scan position 800 as depicted by line 5-5 of Fig. 3; and Fig. 6 is a diagram of a suitable circuit arrangement for hardware that can compact the object image intelligence data.

Detailed Description of Preferred Embodiment The present invention is adapted for use with conveyors or other arrangements where a series of like objects are moved on a repetitive basis for automated inspection or assembly.

The invention serves as a substitute for human inspection of the object shape or orientation and is adapted to provide data representation concerning a part that may have a resolution as little as 0.0005 inches (0.013 mm).

In the illustrated conveyor 10 of Fig. 1, objects 12, 14, 16 rest on a surface 18 that moves in a counterclockwise direction while a tilted central disk rotates at a slower speed to load objects in spaced positions along conveyor surface 18 in a known manner. The objects 12, 14, 16 pass between a camera sensor 22 and a light source 24 after which they move downstream to a conventional detector 26 and diverter 28 which enables reorientation and/or rejection of improperly oriented or sized articles. The diverter may of the general type as show in Dean et al U.S. Patent No.

4,619,356.

As illustrated, a camera sensor 22 is not a raster scan type, but instead consists of a linear array of charge coupled device (CCD) units. The CCD units are aligned to be transverse to the direction of object movement. The linear array of CCD units thus may be essentially vertical in the case of a horizontal conveyor. The CCD units are aligned in a single column to provide a field of view that is one pixel wide and at least about 1000 pixels high. The height of the CCD unit column must be sufficient to span the feature of interest of the object 12, 14, 16 on the conveyor 18. For many small objects such as bolts, screwdriver handles, small caliber ammunition and the like, a maximum variation of the feature of interest may be within a one inch (2.54 cm! span.

Silhouette image data obtained for certain applications must have a 0.0025 inch (0.064 mm) resolution. The number of CCD units in the one inch column may conveniently be about 2000 and advantageously may be 2048. An even smaller resolution below 0.0005 inches may be obtained with the use of about 3000 or 4000 pixels in a one inch column. The linear array of CCD units may be obtained commercially from Texas Instruments as TC-103-1. The drive circuitry necessary for proper CCD operation and timing diagrams to provide a sequential scan of the analog voltage signal are commercially available. The scan rate must provide sufficient time to transfer each pixel charge fully and not allow any charge to accumulate in a pixel between reset and the next scan at which time a momentary voltage is applied to each of the CCD sensing units.

In the system of the present invention the light source 24 is located across the conveyor surface 18 to face the CCD units. As an object 12, 14, 16 passes between the light source 24 and the camera sensor 22, a shadow is formed on certain of the pixel areas whereas unblocked pixels are fully illuminated by the light. By use of a collimated light source which operates through a lens having a shape and size corresponding to that of the linear array of CCD units forming a camera sensor, a precise point on the upper edge surface of the object can be optically determined with great accuracy.

Variations in ambient light conditions are less likely to interfere with operation of the camera sensor when a collimated light source is used.

If the object has a point on the lower edge surface that is positioned above the conveyor surface, a light beam will be detected at appropriately positioned pixels in the same linear array at a point on the lower surface which is opposite the detected point on the upper object surface. Similarly, an aperture in the object which is aligned between collimated light source and the camera sensor will produce transitions in the adjacent pixels to provide a manifestation of the marginal edge points of the aperture at successive positions as the object advances past the camera sensor.

Successive exposures of the camera sensor 22 to each object 12, 14 or 16 as it moves along the conveyor path 18 gives successive data inputs which may be sequentially processed and collectively used to provide as a display, a silhouette of the object before the object reaches the diverter station 28. Object speed on the conveyor may be several inches per second depending upon the desired resolution. For a 2048 pixel CCD unit array, we have found that a scan can be effected in about 330 microseconds and that time periods between successive scans can be variable, with object speeds of up to about 7 inches per second (0.18 ms'l) possible while maintaining a 0.0025 inch resolution.

Successive scans may be provided at 300 microsecond intervals with a 2048 pixel linear array driven by a 10 MHz clock.

Conveyor speeds up to seven inches per second may be acceptable without exceeding the resolution accuracy specified. Object speed on the conveyor may be monitored and signals generated corresponding to object or article displacement along the conveyor path to allow article shape "learning" procedures to be carried out at a speed that is different from the operating speed, or to provide compensation for speed variations generally.

The installation as illustrated in Fig. 1 may include also a system control 30 and control box 32 which are usually physically located near the conveyor, and may be in a single housing.

With reference to Fig. 2, a functional block diagram of the camera sensor 22 is illustrated. The vertical column of CCD units 34, consisting of a 2048 pixel linear array in the illustrated embodiment, is connected to receive clocking or timing signals from the clock and sync circuit 35. Clock circuit 35 includes an oscillator running at a frequency of at least about one MHz, and 10 MHz in the illustrated example, in order to provide pixel scanning in about 200 microseconds and 100 microseconds or more for reset operation. The CCD units that are commercially available are capable of running at clock frequencies as high as 40 MHz. Thus, pixel scan during a 300 microsecond sampling scan after conditioning, is used to produce an analog information signal which contains a transition relating to the precise position of an edge point on an object or part which is being conveyed. To allow for variations in conveyor speeds, the actual start of each vertical slice scan follows receipt of a master reset pulse (Figs. 4 and 5) from microprocessor unit 54 on lead 54 shown in Fig. 2.

From the column of CCD units 34 which each functions as a pixel sensor, an output signal on lead 36 is in the form of an analog signal voltage (see Figs. 4 and 5) containing sequentially obtained voltages of a first amplitude for shadowed pixels and a second low amplitude for those pixels receiving light from light source 24. The analog information is a serial bit stream of uniform length and is transferred serially at the clock rate to a voltage follower that serves as an isolation circuit 38 and to a black sample and hold circuit 40 which produces a voltage level reference signal from pixels that are blocked from receiving light. This provides a reference signal which holds the analog signal at a controlled DC level and may be used as one input to circuitry associated with an analog to digital conversion circuit 42, for control of the sensor output threshold at which the transition in the digital signal occurs.

The output signal on lead 44 is applied to the transition detector and data compaction circuitry 48 which will be described in connection with Fig. 6. On lead 46, a clock signal from the clocking and sync circuit 35 is applied to maintain synchronization between the data compaction unit 48 and the scanning means that is part of the charge coupled device array 34.

The output signals from the data compaction device 48 on leads 50 is in the form of a single binary number for each transition from the analog to digital conversion circuit and is applied to the memory 52 which serves as a buffer to collect all of the data for a particular object 12, 14 or 16 on the conveyor surface on a first in, first out basis. The microprocessor unit 54, which may be any suitable type that is commercially available, may start to process the output signals as soon as the memory 52 begins to receive valid object data.

The camera sensor 22 is thus synchronized with a counter in the data compactor 48 by means of the clocking and sync circuit 35 to provide scan slice information. The memory 52 for data buffering may have a 64K or even smaller capacity for objects of the type mentioned above. As pointed out above, low cost commercially off-shelf available components have a capability to operate up to a 10 MHz data rate in a reliable fashion thereby providing a low cost hardware product.

With reference to Fig. 3, there illustrated a round of ammunition which has a cylindrical cartridge or casing 56 that is supported on a conveyor surface 18 and a projectile 58.

Fig. 4 contains a group of waveforms taken along line 4-4 of Fig. 3 and Fig. 5 contains a group of similar waveforms taken along line 5-5 of Fig. 3. Fig. 4 waveforms are taken at a position corresponding to scan 120 whereas, the Fig. 5 waveforms are taken at scan 800.

In Fig. 4, the waveform of the amplified analog signal starts at time 0 in a black condition because of the conveyor 18. At pixel 30, which corresponds to count 30 in a counter, light is detected thereby starting a negative going digital pulse and a positive going edge detector pulse 60. At pixel 100, the lower edge point on the silhouette of the projectile 58 is effective to block light and create a further edge detector pulse 62. At pixel 500, the light is again detected, thereby causing a third edge detector signal 64 to be generated. Finally, at the top of the sensor array and pixel 2048 of the linear array, the scanner no longer produces a signal and an end of scan transition detector pulse 66 is generated.

A conventional binary counter capable of counting up to at least 2048 at the clock frequency is synchronized with the scan of the 2048 pixels in the camera sensor as indicated at the bottom waveform of Fig. 4. The clock is reset to start at zero as the scan starts so that count values of 30, 100, 00 and 2048 are stored in the memory 52 of Fig. 2 as determined by the time of occurrence of edge detector pulses 60, 62, 64 and 66.

Fig. 5 shows the corresponding waveforms that occur at scan 800. Since the lowest point on the cylindrical casing 56 rests on the conveyor surface 18, the lowest 1499 pixels in the linear array are dark and the first transition occurs with pixel 1500, which is aligned with the upper edge point of the cartridge casing 56 at scan position 800.

The edge detector pulse 68 is generated in response to the transition at pixel 1500 and causes the count value of 1500 to fall through the memory 52 to its output terminals. A similar edge detector pulse 70 occurs at count 2048.

Thereafter, a master reset pulse is generated. The counters are reset to a zero count by a counter reset signal which is synchronized with the beginning of the next scan of the pixels.

Fig. 6 shows one preferred embodiment for converting the digital signals of Figs. 4 and 5 into count values that are supplied to the microprocessor unit (MPU) 54. The digital signal from Fig. 4, in the form of incoming serial binary bits, is applied to terminal 80 of a negative and positive edge detecting network that detects changes in the binary state and issues for each positive or negative edge a 50n sec.

pulse on lead 82. At a 10 MHz clock frequency, the scanned information data and clock counts are separated by 100n sec.

The 50n sec. pulse is used to gate on the memory unit 52 (Fig.

2) which includes FIFO registers 84 as illustrated in Fig. 6.

The three binary counter registers 86 that operate with clock signals on lead 46 are reset by a counter reset signal on lead 88. The count value on leads 50 is constantly presented to the FIFO registers 84. However, the count values are allowed to drop through the FIFO registers 84 only when an edge detector pulse on lead 82 is present In this example, the count values of 30, 100, 150 and 2048 are stored.

When a count value falls through the FIFO registers 84, the FIFO issues an output ready signal to MPU 54 on lead 92.

When the MPU sees an output ready signal, it issues a shift out signal on lead 94 to FIFO registers 84 which releases the count value immediately to the MPU 90. The data at this point is then coded object image intelligence. This handshaking continues throughout the entire scan cycle and sequentially throughout all scans of a object.

As is evident from the foregoing for the scan 120, only four count values are processed and stored rather than 2048 bits of scan information. Other scans such as scan 800 may have only two count values that are processed. The number of scans for a three inch (7.6 cm) object or article may be about 1000. This number may be decreased where less resolution in the horizontal direction is acceptable thereby further reducing the processing time. This compaction of data increases processing speed and reduces memory size requirements without sacrificing resolution of the silhouette image.

While only a single embodiment has been illustrated, other modifications and variations will be apparent to those skilled in this art. The illustrated embodiment has a degree of sophistication which can be simplified for less demanding applications. It is therefore intended that the variations and modifications which fall within the scope of the appended claims and equivalents thereof be covered thereby.

Claims (23)

1. A method of scanning an edge of an object comprising the steps of moving the object relative to a camera sensor comprising a number of sensing elements arranged to view a linear array of pixels extending transverse to the direction of relative movement and spanning the object edge, and sampling each sensing element in succession so as to scan the pixel array.

2. A method of scanning as claimed in claim 1 wherein the sensing elements produce output signals in the form of a serial data stream of known scan length and the method further comprises the step of converting the serial data stream into digital signal pulses having leading or trailing edges related to a point on the object edge.

3. A method of scanning as claimed in claim 2 wherein the serial data stream has an analogue waveform and the method comprises the steps of supplying the serial data stream to an A/D converter and analysing the output signal of the A/D converter to produce said signal pulses at transition points in the A/D converter output signal.

4. A method of scanning as claimed in claim 3, wherein the output of the A/D converter is serial binary data and an output from a sensing element having the object in its field of view is used as a reference signal to control the sensor output threshold at which the transition in the digital signal occurs.

D. A method of scanning as claimed in claim 3 or 4 wherein scanning of the pixel array is synchronised with a clocking circuit, whose output is gated by the signal pulses, thereby to provide count values related to the position of said object edge point within each scan.

6. A method of scanning as claimed in any preceding claim wherein the sensing elements are light sensitive devices, all arranged to face a light source and the object is moved relative to the sensor along a path between the light source and the elements to thereby shade a first portion of the pixels in said linear array while a second portion of said pixels remains fully illuminated, and the method comprises detecting the positions of adjacent pixels where a transition between a shaded and an illuminated pixel occurs.

7. A method of scanning as claimed in any preceding claim, wherein the sensing element are CCD units, the method comprising the steps of: repeatedly applying a momentary voltage to each CCD unit so as to scan the pixel array; and sampling the output of each CCD unit in succession between successive said voltage applications so as to transfer each pixel charge fully and so as not to allow any charge to accumulate between CCD resets and the next scan.

3. A method of scanning as claimed in claims 5 and 7 wherein the clocking circuit is reset following each application of said momentary voltage.

9. A camera sensor for sensing an edge of an object as the object and sensor are moved relative to one another the sensor comprising a number of sensing elements arranged to view a linear array of pixels extending transverse to the direction of relative movement and spanning the object edge, each sensing element being sampled in succession so as to scan the pixel array.

10. A sensor as claimed in claim 9 wherein the sensing elements produce output signals in the form of a serial data stream of known scan length which are supplied to a circuit which converts the serial data stream into digital signal pulses having leading or trailing edges related to a point on the object edge.

11. A sensor as claimed in claim 10 wherein the serial data stream has an analog waveform, and the circuit comprises an A/D converter to which the serial data stream is supplied, the circuit further comprising an analyser to which the output from the A/D converter is supplied, thereby to produce said signal pulses at transition points in the A/D converter output signal.

12. A sensor as claimed in claim 11 effective to sense a number of edge points during each scan of the object, each edge point producing a transition point in the A/D converter output signal, thereby to give a manifestation of the distance between the edge points of said number in the direction of the pixel array.

13. A sensor as claimed in claim 11 or 12 wherein the A/D converter produces serial binary data and an output from a sensing element having the object in its field of view is used to produce a reference signal supplied to the A/D converter and used to control the sensor output threshold at which the transition in the digital signal occurs.

14. A sensor as claimed in claim 11, 12 or 13 comprising a clocking circuit synchronised with scanning of the array and whose output is gated by the signal pulses, thereby to provide count values related to the position of said object edge point within each scan.

15. A sensor as claimed in any of claims 9 - 14 wherein the sensing elements are light-sensitive devices all arranged to face a light source the object being moved in use relative to the sensor along a path between the light source and the elements to thereby shade a first portion of the pixels in the linear array, while a second portion of the pixels remains fully illuminated, the positions of adjacent pixels where a transition between a shaded and an illuminated pixel exists being detected.

16. A sensor as claimed in any of claims 9 - 15 wherein the sensing elements are CCD units and in use a voltage is applied momentarily to each CCD unit so as to scan the pixel array; the output of each CCD unit being sampled in succession between successive said voltage applications so as to transfer each pixel charge fully and so as not to allow any charge to accumulate between CCD resets and the next scan.

17. A sensor as claimed in any of claims 9 - 16 wherein the pixel array comprises at least 2000 pixels per inch (787 pixels per centimetre), thereby to give a resolution of 0.0025 inches (0.064 mm).

18. A sensor as claimed in any of claims 9 - 17 wherein the pixel array comprises at least 1000 pixels.

19. A camera sensor for detecting a linear distance in one direction between marginal edges of an object as the object advances relative to the camera sensor position in a second direction that is transverse to the first direction comprising: at least 1000 sensing elements providing pixels aligned in a single column in a direction parallel to said one direction; means periodically applying a momentary voltage in timed relation to resetting a counter means; means periodically scanning said pixels to provide an output analog information signal containing variations corresponding to a position of two object marginal edge points; and means responsive only to such signal variations for supplying transfer signals to said counting means to provide a count value corresponding to the linear distance between said two edge points during each scanning period.

20. The camera sensor of claim 19 wherein the distance between the marginal edges is not greater than about one inch (2.54 cm), the length of the pixel column is no less than the distance between the marginal edges, the number of pixels is about 2000, the period between successive voltage applications is not greater than about 300 microseconds and the count value represents resolution of about 0.0025 inches (0.064 mm).

21. A camera sensor for determining the position of an object edge relative to a known camera position comprising: a plurality of sensing elements providing pixels having density of at least 1000 per inch (394 per cm) aligned in a single linear column which extends across expected location for the object edge and which faces a collimated light source; and means for identifying which of said pixels is aligned with a point on the object edge which includes: means for scanning said pixels in timed relation with operation of a counter means; an analog to digital conversion circuit connected to receive an analog information signal from said scanning means and to provide an output transition signal corresponding to the object edge point location; and means for transmitting said transition signal to record a value in said counter means that identifies a point along said object edge.

22. The camera sensor of claim 21 including a sync signal generator for effecting successive scans and resetting said counter means on a periodic basis, means for moving the object to successive positions along a path that is transverse to the direction of the linear column and means for storing successive recorded values that are related to object edge profile.

23. The camera sensor of claim 21 or 22 wherein the total number of pixels is about 2000, the time period between successive scans is about 300 microseconds and the resolution in the direction of the linear column is of about 0.0025 inches (0.064 mm).