GB2165120A - Line scanners - Google Patents

Abstract

Line scanners which employ a rotating reflective polygon 32 to sweep an image of a viewed scene 31 across one or more detector 33 suffer from image distortion towards the edges of the field of view. This problem may be mitigated by arranging firstly that the signal output by the detector be sampled, instant by instant, each sample representing a digitized dot-like portion of the viewed scene, and secondly that the sampling be at such a rate as exactly counteracts the linear distortion, so that as the linear distance on the ground increases - as the angular spacing of the instantly viewed area from some central datum line increases - so the sampling rate increases in correspondence (and thus the number of dot-like portions increases to match the greater extent of the scene). In addition the output of the line scanner may be stabilised against lateral angular movement (roll) of the scanner by sampling each line for a time that is shorter than the time needed to scan the maximum field of view so that the scanner output constitutes a window on to that maximum field of view, the initiation time of the sampling for each line being changed thereby moving the window to counteract any roll of the scanner. <IMAGE>

Description

SPECIFICATION Line scanners This invention concerns Line Scanners, and relates in particular to such scanners which employ a rotating reflective polygon to sweep an image of a viewed scene across one or more detector.

The concept of television is now generally well known. In the television camera an image of the scene viewed by the camera is focussed upon a light-sensitive electrically charged target, causing changes in the target's charge distribution depending upon the image's brightness; the target is scanned, a line at a time, by an electron beam, the current flowing along the beam depending upon the charge of the target area on which it is presently impinging; and, either directly or indirectly, there may be derived from this changing current a signal that defines the charge distribution, and hence the viewed scene, and can, in a television receiver, be viewed to reconstitute the image of the viewed scene.

However, one of the earliest television systems - the Baird system - used a quite different technique.

Here, the scene itself was scanned by a spot of light (hence the system was known as the flying spot system), and the light reflected back from the spot as it crossed and recrossed the scene was focussed upon a single detector whose output then represented the brightness of each part of the scene, and was used to generate a picture in the receiver (where, again, there was employed a rotating disc system very different from the present day cathode-ray-tube receiving devices).

There are many things to be said in favour of the flying spot system, and modern technology has found uses for it - or a related arrangement in which most of its disadvantages (which were considerable) have been overcome. One modern version does not scan the scene with a flying spot of light, focussing the viewed light upon a detector, rather it forms an image of the scene, using the available light, and then scans the image itself across the detector.Such an apparatus is known in the Art as a line scanner: an image of the viewed scene is formed (making use of the "natural light" reaching the apparatus from the scene) via a scanner device - usually a reflective polygon that can mechanically be rotated round or oscillated back and forth - that enables the image to be scanned across a "natural light" detector; after each single scan (in which the detector sees a narrow strip, or line, of the image) it is arranged that the image should, either in fact or in effect, be moved orthogonally to the scanner direction, so that by the next scan the detector sees a different strip, adjacent the first, of the image; and the information gathered from the detector's output, strip by strip, is used to construct a picture of the image - and thus of the viewed scene. In detail, such line scanners operate in a number of different ways.Thus: the "natural light" may be radiation of any detectable variety (i.e., visible light or Infra-Red - the detector is naturally chosen to be able to react to the radiation it is required to "see") - and is normally (though not necessarily) that provided by the ambient lighting conditions rather than some artificial illumination; and the orthogonal movement of the image from scan to scan may be caused by a second scan (of the image itself) orthogonal to the first or by simply moving the entire apparatus in such a direction ("orthogonal" to the scan direction) that the next scanned line (part of the new image) relates to a strip of the viewed scene that is orthogonally adjacent to the preceding strip.Such line scanners are described in many published documents, for a whole variety of uses (though most are connected with thermal imaging or terrain scannings systems - either on the ground, as in UK Patent Specification No. 1,317,713, or US Patent Specification No. 3,730,985, or in the air, as in UK Patent Specification No. 1,503,761).

Though for most purposes the line scanners presently described in the Art are quite acceptable, nevertheless they do suffer from image distortion associated with the use of a mechanical scanning device (the rotating or oscillating mirror arrangement) which scans at a constant angular velocity despite the fact that the image is formed in a plane (and is thus flat). The problem is perhaps best appreciated by imagining not that the system is scanning a flat viewed scene image across the detector but that it is scanning the detector across a flat viewed scene (almost as though the detector was a source of light, and the system was a Baird-type flying spot system). The two situations are optically much the same the path of a ray of light between scene and detector is the same regardless of the direction of the ray so it is fair to visualise the matter in the second way.It can then easily be seen that as the scan advances nearer and nearer to the edge of the scene so the distance across the scene covered by the spot becomes greater and greater for every degree of rotation of the scanner mirror, until it becomes infinite as the mirror turns the last fraction that brings the spot-generating light beam exactly parallel with the plane of the viewed scene. Clearly, then, the image is distorted - extended - towards the edges of the field of view, and this may cause quite a problem if, as may well be the case, it is a necessary requirement of the equipment using the line scanner that the output be a completely undistorted representation of the viewed scene (such a requirement would exist, for example, if the equipment was preparing topographically-accurate maps of an area of ground being viewed from above).

This problem of off-center distortion, which gets worse as the field of view widens (and the scan reaches further and further from the centre of the viewed scene), may at least be mitigated, according to the present invention, by arranging firstly that the signal output by the detector be sampled, instant by instant, each sample representing a digitized dot-like portion of the viewed scene, and secondly that the sampling be at such a rate as exactly counteracts the linear distortion, so that as the linear distance on the ground increases - as the angular spacing of the instantly viewed area from some central datum line increases - so the sampling rate increases in correspondence (and thus the number of dot-like portions increases to match the greater extent of the scene).

In one aspect, therefore, the invention provides a method of counteracting the viewed scene distortion observed when using line scanners as the angular spacing of the viewed area from some central datum line in the viewed scene increases, in which method a) the scanner's detector output is sampled, and b) the sampling is effected at a rate which is varied in correspondence with the expected distortion.

In another aspect the invention provides apparatus for counteracting the viewed scene distortion observed when using line scanners as the angular spacing of the viewed area from some central datum line in the viewed scene increases, which apparatus comprises: a) means for sampling the scanner detector's output; and b) means for effecting the sampling at a rate which is varied in correspondence with the angular spacing, and thus with, and counteracting, the expected distortion.

The invention provides a method and apparatus for counteracting viewed scene distortion in line scanners. These scanners are generally of the kind described hereinbefore. Specifically, such a scanner employs a rotating or oscillating mirror - usually in the form of a reflective polygonal (often hexagonal) prism - which rotates, or oscillates, at a constant angular rate. Moreover, the scanner may be designed and will then use a suitable detector - for imaging a scene in any appropriate part of the electromagnetic spectrum, two typically varieties using respectively Infra-Red and visible light.

The viewed scene distortion the invention counteracts is that distortion caused by scanning a flat image of the viewed scene across the detector at a constant angular rate rather than a constant linear rate.

If the detector output is then digitised (to provide a train of digital signais each defining the detector output level at some instant in time) at a constant rate (in line with the constant angular scan rate) the signals representing points at or near the edge of the field of view will necessarily be spaced further apart linearly (though equally apart angularly), and if used to construct a video picture (where each digital signal is a pixel, and the linear spacing of adjacent pixels is the same across the entire width of the picture) the effect will be to squeeze the picture up at or near the edges. As is explained in more detail hereinafter, the amount of this linear distortion is dependent upon the angular spacing of the viewed area from some central datum line (conveniently the centre of the field of view); the further to either side the viewed area the greater the distortion.

The distortion the invention seeks to correct arises because the line-scanning-using equipment constructs a picture (or some other record) of the viewed scene from samples of the scanner's output, and these samples are taken at a constant rate despite the fact that the image is being swept past the scanner detector at a rate which, though angularly constant, is not linearly constant. The proposed way of correcting for this is to adjust the sampling rate, from moment to moment, so that the change in the number of samples per unit (time or angle) exactly counteracts the distortion otherwise observed - thus, that the sampling rate is now constant relative to the linear dimension of the image, and the same number of samples is taken per unit length of image regardiess of whether the viewed area is central or peripheral of the viewed scene.Specifically, the sampling rate is increased as the angular spacing of the viewed area from the chosen central datum line increases. For scanners using a detector having individually a small field of view (a narrow angular subtense, as is usually the case) the length of the viewed area is proportional to sex26 where () is the angular spacing of the area from the central datum line, and this area advances across the detector at a linear rate proportional to tan e - i.e., both the viewed area and the distance between adjacent viewed areas (determined by the sampling rate) are non-constant. If, then, the sampling rate is adjusted in exact correspondence, the between-viewed-area spacing can be made constant, and the linear distortion removed.Specifically, if the sampling rate is adjusted by a factor proportional to tan (3 - so that as the spacing increases so the sampling increases - then the sample rate per unit length (of viewed scene) becomes constant, and the linear distortion of any picture (or other representation) derived from the scanner's output is removed.

There are a number of ways in which the sampling rate can be arranged to vary in the required manner as the viewed area's angular spacing varies.

However, because for any given system it will be known how many samples are required per total line of scan, and the calculations to determine how these should be allocated - how the sampling rate should vary - as the angular spacing changes can all be done in advance, it is very convenient to store a series of numbers representing a series of sampling rates as the spacing changes, and to call up, and apply, the relevant number (and thus sampling rate) as required. For instance, if it were the case that for angular spacings of 1, 2, 3, 4, 5....10 degrees the relevant sampling rates were 2, 3, 5, 7, 9....17 then these numbers could be electronically stored in "boxes" identified by the spacing value, and each extracted and used for the relevant spacing. Alternatively, and as effected in a preferred embodiment, there is stored a number representing the reciprocal of the sampling rate -i.e., a number related to the time gap between samples, which gap decreases as the sampling rate increases. In practice, such an arrangement can be employed in the following way. Utilising a memory chip (a Programmable Read Only Memory, or PROM) there may be stored the time gap between samples for all likely angular spacings; in use, the system electronics counts the number of master clock pulses generated by the scanner (by the section driving its rotating polygonal reflector, say, and so linked to the angular spacing), and itself generates one "sampie" pulse after counting each defined number of master clock pulses, at the same time advancing the PROM location (via an address generator) to the next gap number.In this way the length of time between samples decreases - and so the sample rate increases - as the angular spacing increases.

The ability to change the sampling rate according to the invention can optionally be associated with an additional feature allowing the field of view to be stabilised against sideways angular movement of the platform carrying the scanner. For example, one particular use of line scanners is in an aircraft, where the scanner looks down to the ground just in front of, or even directly underneath, the aircraft, and it will be usual for the scanner apparatus to be fixed (on a platform, perhaps) relative to the aircraft so that if the aircraft rolls (or banks) to left or right then the scene viewed by the scanner naturally shifts right or left by some corresponding amount.This may in fact be most undesirable - it may be necessary to look at the same angle in front of or beneath the aircraft no matter how (within reason) the aircraft rolls - and a means of at least partially mitigating the effects of roll when employing a system where the scanned view is narrower than the overall field of view - as is generally the case - is to arrange that the time of initiation of each line scan is altered to reflect, and counteract, the change in the field of view caused by aircraft roll. Within limits, then, the initation of a line scan may be advanced (made earlier) if the aircraft (or other platform carrying the scanner) rolls so as to move the scanner view in the direction of the scan, or may be delayed (made later) if the aircraft (or platform) rolls so as to move the scanned view in the direction opposite to the scan.The effect of this is as though the scanned area constitutes a window smaller than but looking onto the total possible field of view, and this window can be moved to one side or the other so as to compensate for (small) lateral changes in the total field of view caused by rolling/ banking of the scanner platform.

This concept of "stabilising" the scanner-derived data against roll of the scanner by suitable altering the initiation time of each scan is itself novel and inventive. In another aspect, therefore, the invention provides a method of stabilising the output of a line scanner against lateral angular movement (roll) of the scanner, in which method a) the scanner's detector output is sampled, b) the sampling for each line is effected for a time that is shorter than the time needed to scan the maximum field of view, so that the scanner output constitutes a window onto that maximum field of view, and c) the initiation time of the sampling for each line is changed, so moving the window, to counteract any roll of the scanner.

The time during which the scanner detector's output is sampled is shorter than the time necessary to scan the maximum field of view (which field is provided by the optics of the scanner via, say, the rotating reflective polygonal prism). Just how much shorter it is depends upon individual requirements (how wide the maximum field of view is, and how wide the scan needs to be). To give some idea, however, it will commonly be the case that the maximum field of view is about 1200 wide, while the scanned view is 80" wide, so that the scanned view (the window) - normally centrally located within the maximum view available either side - can be moved up to 200 in either direction, so compensating for up to + 200 of roll.

Initiation of scan sampling is in practice quite simple. A typical system will use a master clock to drive a counter, the counter outputting a "sample now" pulse each time a specified number of clock pulses has been received, and it is an easy matter to inhibit pulse output for a specified length of time. And, of course, this length of time may be linked to a scanner platform roll detector, so that it changes depending upon the angle of bank (roll) of the scanner.

The invention extends, of course, to a line scanner whenever using the distortion- or roll-counteracting method or apparatus of the invention as described and claimed herein.

Various embodiments of the invention are now described, though only by way of illustration, with reference to the accompanying Drawings in which: Figure 1 is an artistic representation of the major elements of one type of equipment using a line scanner; Figure 2 is a sectional view of a line scanner as viewed in the equipment of Figure 1; Figure 3 is a diagrammatic view of a line scanner showing the essential principles of operation; Figure 4A and B are diagrammatic views showing further the principles of operation, and the problems, of a line scanner; Figure 5 is another diagrammatic view explaining the benefits of one embodiment of the invention; Figure 6 is a general block diagram of circuitry useable to apply the invention to a line scanner.

Figure 1 shows the main components of a line scanner system carried by an aircraft (not shown) and employed to map from the air the terrain over which the aircraft is flying. The scanner itself (11; shown in a more detailed sectional view in Figure 2) receives radiant energy (in this case, Infra-Red) along a split path (A, B; A is shown slightly stippled) from the terrain (generally T) below the aircraft, and via a system of a rotating reflective tetragonal prism (21), a number of plane mirrors (22A, 23A and 22B, 23B), and a concave, focussing mirror (24), directs this energy onto a detector (25).The output from the scanner 11 (which (is "scanning" the terrain T in a swathe extending to left and right of the aircraft's track, successive scans, as S1, S2 and S3, being adjacent one another because of the aircraft's forward motion F), is fed to the conversion and sampling electronics (12), which in turn feed a signal to a Cathode Ray Tube (CRT; 13) where they are used to construct a line across the face (not shown) of the tube that represents a visible image of the terrain below scanned in Infra-Red.Naturally, successive scans would result in successive lines on the CRT 13, so building up a picture (as on a television) of the ground over which the aircraft is flying, but in this embodiment the single scan is focused (at 14) onto a length of photographic film (15), which moves on at a speed related to the velocity-to-height ratio of the aircraft, so that there is formed a strip map record of the overflown terrain.

Figure 3 shows, in very simplified form, the scanning of an extended target (31; in Figure 1 this is, of course, a strip of ground, as S1, S2 etc.), with the mechanical element of the scanner (the reflective prism 21 in Figure 2) represented by a plane mirror (32; shown in three separate positions, LM, CM and RM. The effective instantaneous field of view of the mirror 32 has a very narrow angle, so that the area of target it sees (as Lr Mr and Err), and reflects onto the detector (33) via a focusing system (here shown as a lens 34) is correspondingly very small. The situation may best be understood by imagining that the mirror is, instead, projecting onto the target an image of the detector (the radiant energy pathways are identical no matter which way the energy travels!).The actual instantaneous field of view of the mirror is, of course, somewhat wider, but only a small portion of it is focused onto the detector as the energy from the target is "swept" passed the focusing system (the idea of "sweeping" the image passed the detector is represented, not strictly accurately, by the three overlapping target "images" produced in the detector plane at L,, C, and R, (the images are, of course, mirror images, and thus left-right reversed).

Visualising the scanner in terms of the sweeping of an image past the detector is not easy, and a clearer understanding of equipment, it's problems, and the solutions provided by the invention can better be gained by visualising instead the converse - that is, the sweeping of a projection of the detector across the target. The latter viewpoint is used hereinafter.

Figures 4A and 4B help to explain the problem of growing line scanner image distortion as the scan extends further from a centre position (immediately under the aircraft, say). In both Figures the box 41 represents the scanner, the heavy line 42 the scanned area, and the light lines joining the two represent ray paths (R) for instantaneous projections of an image (I) of the scanner's detector (not shown separately) onto the ground. In Figure 4A there are shown ray paths (Rgo etna!) at successive 50 angles starting from the verticai (thus, working to the left as viewed the angles between the ray path and the - flat target area are 90, 85, 80 55, 50).In a real system, of course, the scanner output might by sampled at a rate corresponding to an angular spacing as little as 0.1 . or even less.

It is evident from Figure 4A that for equi-angular separation of ray paths R the spacing of successive projected detector images / on the target area increases as the angular spacing of the ray path from the central (vertical) line increases. If, then, each detector image / were now to generate a pixel (P) on a television tube surface (represented by the heavy line 43), the between-pixel spacing being constant across the surface, the resulting image of the target area would be distorted - squeezed up - the distortion becoming significantiy worse towards the edges of the picture.

A mathematical approach to the same situation is now given with reference to Figure 4B.

The scanner 41 is located at height h above the plane of the target area 42. The scanner sightline - the raypath of the projected detector image - is at an angle Oto the normal (44) to the plane. The scanner angular subtense m (in reality the projector has a finite, but small, angular width) projects as the footprint x at a distance Xfrom the nadir onto the plane. These parameters are related by the equations:

and X = h tan (3 (2) The footprint length is then proportional to sex2(3. A scanner mirror rotating at a uniform speed will then sweep out equal angles (3 with each equal increment of time, thus advancing the footprint by unequal increments proportional to tan 0.

Conventionally, the analogue output signal is digitised by sampling at periodic intervals (by means of a master clock) so as to divide the line into a number of picture elements (or pixels) of equal time duration (or period). Each pixel then usually corresponds to the detector angular subtense. If this sampled signal is then used to form a picture, by displaying each pixel contiguously, the resulting picture will have along-scan distortion due to the unequal spatial sampling.

As is explained herein, the invention corrects this distortion by advancing the footprint by equal spatial intervals along the plane of observation. This is achieved - as described hereinafter - by the generation of a non-linear pixel sampling clock that is synchronised to the scanner line period to fit in a preset number of pixels within a line of varying duration.

A second aspect of the invention enables there to be mitigated the effect of a limited sideways angular movement of the scanner equipment - more particularly, of the carrier for the equipment, as in the case of an aircraft rolling or banking to one side. Such a roll/bank situation is depicted in Figure 5, where the scanner is shown in two orientations (41b in solid line and 41v in dashed iine). It is imagined that the scanner (41b) is banked to the right by a small angle 0 such that its "straight down" view (along the solid line 54) is actually at an angle W to the true vertical 53 (shown dashed).It is further imagined that the scanner has an actual field of view (the total angular scan) bounded by the solid lines A,, AR, and symetrically disposed about the line 54, but is electronically constrained to view a more limited field (the active scan window) bounded by the solid lines CL, CR also symetrically disposed about the line 54. Such a situation can arise because the start of sampling of the scanner's detectors output is deliberately delayed until some time after the start of an actual scan, and the end of sampling is similarly deliberately advanced to some time before the end of an actual scan.Under these circumstance, the more limited field of view (the scan window) can be moved to the right - to that which would be seen by the scanner in its unbanked state 41v, shown bounded by the dashed lines UL, UR - by the simple expedient of correspondingly changing the sampling initiation delay and the sampling end advancement. Thus, if the target area 42 is being scanned from left to right (in the direction of the arrow), and the equipment banks to the right (as shown), the sampling initiation delay is increased, and the sampling end advance decreased, to compensate for the bank, and to make the scanned area the same as would otherwise be scanned if the equipment were not banked, so centering the scan about the nadir.

Naturally, this technique cannot be used to compensate for severe roll/bank, because within the actual field of view (bounded by the lines AL, AR) there is only so much leeway in sampling initiation and ending times. Other means - such as gyrostabilising the scanner equipment - may, however, be of use in such extreme circumstances, as is well-known in the Art.

Finally, Figure 6 is a block circuit diagram showing the details of one embodiment of the electronics needed to put the invention into effect. The broad outlines of the circuitry are revealed here, for the Figure shows how constant rate clock pulses (67) representing the constant rate angular scanning position of the scanner are used to drive a counter (61) that outputs varying rate pulses (60) that drive the detector sampling electronics (not shown). The system operates to generate a non-linear pixel sampling clock in the following manner. A master clock (67) is frequency- and phase- locked to the scanner's polygon rotational-speed signal (68) within a phase lock loop (69). The non-linear pixel clock (60) is then produced by counting a pre-set number of master clock pulses in the pixel counter (61), and outputting a pulse at the completion of the count. The count size for each consecutive output non-linear pixel clock pulse is stored in a pre-determined sequence in a PROM (62) which is addressed by an address counter (63). The address counter is incremented each time a non-linear pixel clock pulse is output.

The total number of master clock pulses contained within the total angular line scan is constant, determined by the phase lock loop multiplier and by the scan efficiency, and so the non-linear pixel sampling clock is fully adaptive to varying scan rates, for a fixed number of unequal-duration clock pulses is fitted exactly within the total angular line scan period.

To move the scan window within the total scan, so compensating for roll, a polygon-position pulse 64 initiates the pixel counter via an adjustable roll delay counter 65 to locate the active line scan within the total angular line scan.

Claims (15)

1. Apparatus for counteracting the viewed scene distortion observed when using line scanners as the angular spacing of the viewed area from some central datum line in the viewed scene increases, which apparatus comprises: a) means for sampling the scanner detector's output; and b) means for effecting the sampling at a rate which is varied in correspondence with the angular spacing, and thus with, and counteracting, the expected distortion.

2. Apparatus as claimed in Claim 1, wherein the scanner employs a rotating or oscillating mirror to scan the viewed scene image across a detector.

3. Apparatus as claimed in either of the preceding Claims, wherein the sampling rate is adjusted by a factor proportional to tan o (where (3 is the angular spacing of the area presently being viewed by the detector from a chosen central datum line).

4. Apparatus as claimed in any of the preceding Claims, wherein the sampling rate control means includes means for storing a number representing the reciprocal of the sampling rate for all likely angular spacings; and means for counting the number of master clock pulses generated by the scanner, for generating one "sample" pulse after counting that number of master clock pulses as defined by the sampling rate reciprocal number, and for then selecting the next sampling rate reciprocal number.

5. Apparatus as claimed in any of the preceding Claims, wherein the sampling for each line is effected for a time that is shorter than the time needed to scan the maximum field of view, so that the scanner output consitutes a window onto that maximum field of view, and there are means to change the initiation time of the sampling for each line, so moving the window, to counteract any lateral angular movement of the scanner.

6. Apparatus as claimed in Claim 5, wherein, where the maximum field of view is 120 wide, the scanned view is 80 wide, so that the scanned view (the window) - normally centrally located within the maximum view, with 20 available either side - can be moved up to 20 in either direction, so compensating for up to + 20 of roll.

7. Apparatus as claimed in either of Claims 5 and 6, wherein a master clock drives a counter, the counter outputting a "sample now" pulse each time a specified number of clock pulses has been received, and there are means to inhibit pulse output for a specified length of time.

8. Apparatus as claimed in any of the preceding Claims and substantially as described hereinbefore.

9. A line scanner whenever using the distortion-counteracting apparatus as claimed in any of the preceding Claims.

10. A method of counteracting the viewed scene distortion observed when using line scanners as the angular spacing of the viewed area from some central datum line in the viewed scene increases, in which method a) the scanner's detector output is sampled, and b) the sampling is effected at a rate which is varied in correspondence with the expected distortion.

11. A method as claimed in Claim 10 and substantially as described hereinbefore.

12. A line scanner whenever using the distortion-counteracting method as claimed in either of Claims 10 and 11.

13. A method of stabilising the output of a line scanner against lateral angular movement (roll) of the scanner, in which method a) the scanner's detector output is sampled, b) the sampling for each line is effected for a time that is shorter than the time needed to scan the maximum field of view, so that the scanner output constitutes a window onto that maximum field of view, and c) the initiation time of the sampling for each line is changed, so moving the window, to counteract any roll of the scanner.

14. A method as claimed in Claim 13 and substantially as described hereinbefore.

15. A line scanner whenever using the roll-counteracting method as claimed in either of Claims 13 and 14.