GB2037119A - Reducing the effect of defective elements in a thermal image detector array - Google Patents

Abstract

In thermal imaging systems in which a line array of thermal radiation detectors 3 is scanned transversely to its length relative to an image of a thermal scene to produce a visible reconstruction of the thermal scene having as many picture lines as detectors in the array, the variations of detector gain and noise along the array 3 and failures of detectors produce "liney" pictures and absent lines respectively. To reduce these effects a scanning cycle is used in which, between scans, the thermal image 17 is displaced along the array by a multiple of the detector pitch. In a group of such scans 18, 19, 20, 21, each line in the picture is scanned in turn by several different detectors. The visible image is reconstructed by methods which average the signals of the different detectors for each line, for example by persistence of vision, reducing the degrading effect on the picture which any one detector can have by virtue of excess noise or zero signal. <IMAGE>

Description

SPECIFICATION Thermal imaging systems This invention relates to thermal imaging systems comprising an infra-red optical system for focussing a thermal image of a field of view upon a linear array of thermal radiation detectors, a scan generator for scanning said thermal image relative to the array in a repeated scanning cycle, and means for reconstructing a visible version of the field of view from signals supplied by the detectors. More particularly, the invention relates to thermal imaging systems in which redundancy is employed in the scanning cycle to reduce the degradation of the visible picture caused by the variation in gain and noise amongst the detectors in the array and to reduce the effect of a detector which has failed.

A thermal imaging system is known from British Patent Specification No. 1,361,144 in which the scan generator produces a conventional raster of horizontal lines having a vertical frame scanning motion. A linear array of detectors is arranged parallel to the line direction and each line in the thermal image is scanned by all the detectors in the array. The output signals from each detector are delayed as a function of the scan rate and the relative position of the element in the array to allow the summation of signals from the same thermal image elements provided by each of the detectors in the array. Thus the signals of N detectors are added directly but the N noise signals are added root-mean-square-wise to produce an improvement in signal to noise ratio of N.The picture scanning rate however is no greater than it would have been with one detector, although all lines are scanned by the same combined detector and the loss of a detector degrades all lines equally. To achieve either a better a scanning rate or to remove the need for a frame scanning motion which approximates one dimension of the picture, it has been proposed to use a linear array of detectors, equal in number to the number of lines desired in the picture, and to scan this array in a transverse direction only, dispensing with the frame scanning motion. Such systems are reviewed in the text-book "Thermal Imaging Systems" by J. M. Lloyd. Plenum Press 1 975 Chaper 8.Since each detector is solely responsible for the content of the one associated line in the picture, the variations in gain and noise amongst the detectors, and amongst the separate amplifying channels associated with each detector, produce visible "variations from line to line" in the picture which is at least annoying and may interfere with the correct interpretation of picture detail especially if carried out in haste. Also the loss of a detector channel will produce the total loss of information on one line in the picture.

It is an object of the invention to reduce the visible effect of channel variations and loss in thermal imaging systems in which a linear array of detectors is scanned transversely relative to the thermal image.

According to the present invention there is provided a thermal imaging system comprising an infra-red optical system for focussing a thermal image of a field of view upon a linear array of thermal radiation detectors, a scan generator for scanning said thermal image relative to the array in a repeated scanning cycle, and means for reconstructing a visible version of the field of view from signals supplied by the detectors, characterised in that the detectors are regularly pitched along the detector array and in that the scan generator comprises means for scanning the thermal image transversely relative to the length of the array in such a way that the scanning cycle comprises a group of parallel scans in which the thermal image is regularly displaced between adjacent scans along the array length by a multiple of the detector pitch.

If succeeding scans of the group are made with equal intervening displacements in the same direction throughout the group of scans, the visible picture may exhibit an obtrusive vertical crawl of the line structure of the picture. To reduce this crawl effect the scans of the group may be executed in a pseudorandom sequence such that each scan of the group is made once only and in the same relative order in each group, but that the order of the scans as a function of displacement of the thermal image along the length of the array within the group constitutes a random pattern. Off all the possible random sequences of scan positions that are possible once a given number of displacements within the group has been chosen, those are preferred in which there is the least proportion of systematic movement of the scan throughout the group in one direction only.As far as possible, each displacement in the group should be followed by a displacement of roughly equal amounts but in the reverse direction.

The invention will now be explained and described, by way of example, with reference to the accompanying drawings in which: Figure 1 shows a thermal imaging system in block diagram form to which the invention can be applied, Figure 2 shows a first scanning cycle which can be used to embody the invention in the apparatus of Fig. 1, and Figures 3 and 4 show two further scanning cycles.

In Fig. 1, the thermal imaging system is shown in elevation in schematic block diagram form for the purpose of showing an example of how a thermal image may be dissected by an array of scanning infra-red detectors and how a visible image correspond ing to the thermal image may be reconstructed. The infra-red optical system in this case comprises an afocal infra-red telescope 1 and an objective lens 2. The afocal telescope 1 does not itself form a real image of the scene, in thermal radiation, upon the detector array, but functions as a Galilean telescope providing either (a) magnification with reduced field of view or (b) minification with increased field of view. The telescope 1 may be removed and the scene viewed at unity magnification.The virtue of the telescope 1 in this case is that changes of system magnification may be obtained while maintaining parallel radiation between telescope 1 and objective 2, where the scan generator is placed.

The objective 2 forms a real image of the scene, in thermal radiation, upon a linear array of detectors 3, shown schematically only as four detectors. In practice the length of the array is chosen to cover the desired picture height and the number of detectors in the array are chosen to provide the desired degree of vertical resolution. To reduce the number of detectors required achieve a given picture height at a given resolution, the scanning may be arranged in bands, the thermal image being displaced relative to the array in the direction of its length between scans by the length of the array. For example, a 64 element array may be scanned in two bands to produce a 1 28 line picture or in four bands to produce a 256 line picture. Mirror drum scan generators will be described later adapted for such banded scanning.

A scan generator for scanning the real thermal image relative to the array in a repeated scanning cycle is provided by a pyramidal faceted mirror drum 5 rotated by means (not shown) about the axis 8 of the drum 5 arranged parallel to the optical axis of the objective lens 2. The size of each mirror facet 6 is chosen so that a substantial portion of the radiation beam 7 is reflected into objective 2 throughout the useful portion of the scan provided by each facet. Each facet is set at an angle of 45 relative to axis 8. During rotation of the drum 5 each facet 6, as it performs a scan, moves at right angles to the plane of the drawing, effectively rotating about about an axis 9 in its own plane passing through the drum axis 8.The beam 7 is consequently rotated about a horizontal axis in the plane of the drawing and the image focussed on the detectors 3 is moved normal to the plane of the drawing and transversely relative to the length of the array to perform the scan. The scanning cycle in this case is a simple repeated scanning of the same line in the thermal image by the same detector. The focussed image also rotates to a small extent about the optical axis of ojective 2, due to facet motion, curving the scan slightly. It will be seen later that the use of the same drum for image reconstruction compensates for this curvature of the scan.The detectors 3, which typically are cooled cadmium mercury telluride photoconductive detectors responding to radiation in the wavelength range 8 to 1 3 microns, are regularly pitched along the array and are mounted upon a cooling device 4 which, for example, may employ liquid nitrogen or may be a thermoelectric cooling device. Each detector is connected to its own low noise pre-amplifier 10 feeding an associated output amplifier 11 which together constitute a channel amplifier 1 2. Only one channel amplifier is shown for clarity, there being a channel amplifier for each detector.Each channel amplifier feeds a separate modulateable light source, for example a light emitting diode, in the array of light sources 1 3. Each connected detector and light source occupy corresponding positions along their respective arrays. The channels have nominally equal gains, measured as the ratio of light output change from the source for a given change of infra-red radiation incident upon the associated detector. In the transition between succeeding scans the objective 2 receives raditation reflected from two adjcanet facets 6 which are then positioned to reflect radiation from predetermined portions of the interior of the system casing which provide a reference level of radiation for the detectors. The channel amplifiers are a.c. amplifiers having a d.c.

level clamping circuit in the output stage which comes into operation in the reference period between scans and sets the brightness of each source to a common predetermined level which may be adjustable and may function as a picture brightness control. The channel gains may be either all positive so that hot objects are reconstructed as bright objects, or all negative so that hot objects are seen as dark areas. The latter negative gain mode may be useful if low level detail is to be viewed in a scene having hot point sources which would overload the display in the positive gain mode.

The visible picture is reconstructed for viewing by eye 1 5 using persistence of vision. A lens 1 6 applies suitable magnification to the source array 1 3 which is viewed as a virtual image in the facets of the drum 5 via an eye piece 14. In the example of Figl 1, the drum 5 has an even number of facets regularly spaced around the drum, all facets being inclined at 45 to the drum axis. Thus while a given facet on the drum is making a detector scan, the diametrically opposite facet is making a corresponding scan of the sources 1 3.

The virtual image of each source is drawn out into a slightly curved horizontal line modulated in brightness along its length corresponding to the details scanned in a corresponding line of the thermal image by the associated detector. The scan curvatures on the infra red and visible images are equal and compensate. If, for exaple, a drum having 10 facets is rotated at 1 50 r.p.m., the scanning rate is 25 per second which will provide adequately low flicker in the reconstructed picture provided the average brightness of the light sources is not too great.

Thermal imaging systems typified by that described with reference to Fig. 1 have known disadvantages. The responsivities of the detectors in a typical array as presently available are not equal and may exhibit a spread in value of as much as 3 to 1. Also, the noise figures of the detectors exhibit a spread in value which does not necessarily correlate with the spread in responsivities. Thus, there is a spread in the signal-to-noise ratios of the detectors. With presently available amplifying devices, the noise contribution of the channels can be made insignificant. Channel amplifiers can generally be made in batches with a spread in gains of only a few per cent. The light sources in array 1 3 may have a significant spread in luminous efficiency.The combined effect of the spreads in detectors, channel amplifiers and sources may be partially compensated by providing each channel amplifier with a preset gain control which may be adjusted, as a manufacturing operation, to equalise all channel gains. If this is done, the noise levels on the lines in the displayed picture are not equal since each line is the product of the channel. Thus a "liney" picture results which may interfere with the interpretation of scene details. Alternatively, if the channel gains are adjusted to equalise the noise outputs, an equally objectionable "liney" picture is obtained due to the line-to-line differences in gain. In either event, the gain setting operation will be time consuming, will require relatively skilled labour and thus be expensive. This gain setting just be repeated in the event that the detector array has to be replaced.

Another disadvantage of the system of Fig.

1 is that the failure of any part of one channel will result in a "dead" line in the picture. This "dead" line may be permanently black, at a fixed illumination or excessively noisy and, apart from the visual annoyance, may result in a serious loss of information.

Fig. 2 shows schematically a scanning cycle of a system made in accordance with the invention, which scanning cycle comprises a group of four parallel scans in which the thermal image is regularly displaced between adjacent scans along the array length by the detector pitch. In essence, there is a stepdown of the scan between scans, referred to herein as step-down scanning. An embodiment of step-down scanning will be described later. The background 1 7 of Fig. 2 is to be taken as the thermal image, considered for ease of explanation as stationary. The column 1 8 represents the array of detectors 3 in position relative to the thermal image at the start of the first scan of the cycle in the transverse direction 23.In this example there are 1 6 detectors of equal size in a close packed array and hence regularly pitched along the array. Column 1 9 represents the array of detectors in position relative to the thermal image for the start of the second scan of the cycle. Column 1 9 is shown displaced to the right of column 18 for clarity only. In practice, the second scan and succeeding scans start at the same relative position, in the scan direction, as the first scan. Column 1 9 shows the detector array 3 displaced along its length by one detector pitch relative to the thermal image. Thus line 1 of the first scan is not scanned, line 2 of the first scan is now scanned by detector 1, line 3 by detector 2 and so on to new line 1 7 scanned by detector 16.Column 20 shows the third scan displaced by one detector pitch relative to the second scan and column 21 the fourth scan similarly displaced. Column 24 shows the line numbers of the scanned scene. In all, 1 9 lines in the scene are scanned as shown in column 24. Column 22 shows the number of different detectors scanning each line of the scene.

Lines 1 and 1 9 (column 24) are only scanned once in the cycle, there being three absent scans and one active scan. Lines 2 and 1 8 are scanned by two different detectors each, lines 3 and 1 7 by three different detectors and lines 4 to 1 6 inclusive are scanned by four detectors in the cycle with no absent scans, lines 1, 2, 3, 17, 18 and 19 may be masked off if desired. Fig. 2 also represents the corresponding displacements of the scans of the array of light sources during visible picture reconstruction, a line in a given position in the thermal image always being reconstructed by a line in a corresponding position in the visiible picture on all four scans of the cycle.

If the scan rate is high enough for the four scans of the cycle to be effectively added by persistence of vision, the noise and the signal in each of the visible lines 4 to 1 6 inclusive are then the sum of the outputs of four different channels. Thus any two adjacent visible lines, for example lines 8 and 9 (column 24), have three channels in common, channels 6, 7 and 8 in this example, and will differ considerably less in noise and signal, on average, than if each line had been scanned only by its own detector. Differences in channel gain and noise can now be tolerated to a greater extent, with the possiblity that the gain setting operation can be simplified or even avoided altogether.

The failure of any one channel is not catastrophic. For example, the failure of channel 7 in Fig. 2 means that lines 7, 8, 9 and 10 (column 24) are reconstructed by 3 live scans and one dead scan each. Flicker at the scanning cycle repetition rate will be observed on these lines, but the information of these lines will still be present. Up to three adjacent channels can fail before the information is lost entirely, although the flicker would be more pronounced if the scanning cycle repetition rate is below the flicker perception threshold.

At a scanning rate of 100 per second, however, flicker would be at 25 Hz. Also lines reconstructed from dead channels may be of reduced brightness or contrast or may be noisy depending on the nature of the channel fault. Clearly, however, several isolated dead channels can be tolerated before the imaging system becomes unserviceable.

In the description of Fig. 2 for simplicity the order in time in which the four scans occur was assumed to be as drawn, that is 18, 19, 20 and 21. This progressive movement of the array relative to the thermal image in the course of the scanning cycle may give rise to a visible "crawl" downwards of the line structure of the picture. The order of the scans within the scanning cycle may be changed, as will be described later, to a pseudo-random order to reduce this crawl. With only four scans in the cycle there are only two basically different pseudo-random orders typified by 18, 20, 21, 19, 18 etc and by 18, 21, 19, 20, 1 8 etc, all other possible orders being either inversions or phase shifted versions of these two basic orders.

Pictures having greater number of lines may be obtained from a given number of detectors by the use of banded scanning. To achieve banded scanning, the scan generator may comprise means for displacing the thermal image relative to the array along its length by an integral multiple of the length of the array between succeeding scans of the group, means for then performing a scan, and means for then restoring the thermal image to its position in the group relative to the array appropriate to the following scan of the group. Fig. 3 shows the step-down scanning cycle combined with banded scanning. For direct comparison with Fig. 2, two bands of only eight detectors are employed to provide the same total of 1 3 lines, each scanned four times in succession by different detectors. The same drawing conventions are employed as in Fig. 2.Column 25 shows the start of the scan of the top band of the scene and column 26 the start of the bottom band. The pairs of columns 27, 28; 29, 30 and 31, 32 show the starts of the three other stepped-down scans which complete a scanning cycle having eight scans. The scanning rate is doubled as compared with that of Fig. 2 if an equal cycle scanning rate is desired. As in Fig. 2 each line is scanned and reconstructed by four different channels. In the overlap region in the middle lines of the picture, channels from opposite ends of the array contribute to the scanning cycle for a given line, for example the 10th line (column) is scanned by channels 2, 1, 8 and 7.

Fig. 4 shows step-down scanning applied to an array of detectors (and sources), each of which is spaced apart from its neighbour by a detector width. Columns 33 and 34 show the starts of two interlaced scans covering a total of 1 6 lines with 8 detectors. Columns 35 to 40 show the remaining three stepped-down interlaced scans. In Figs. 3 and 4 a single dead channel affects 8 different lines 'who groups of 4 in Fig. 3 and eight adjacent lines in Fig. 4. Also in Fig. 4 only 10 lines are scanned by four different channels each. It must be emphasised, however, that Figs. 2, 3 and 4 show far fewer channels for clarity than would be used in practice.Typically 64 channels may be used, the number of partially scanned lines at the top and bottom of the picture being a function only of the number of different detectors used to scan one line, which will be called the overscan factor hereinafter, and the detector spacing factor if interlaced scanning is used.

The parameters of step down scanning cycles and displayed pictures are related as follows.

if D = number of channels in the array, R = the overscan factor, C = the number of scans in a complete cycle, and F = number of fully overscanned lines, Then for close packed detectors, scanned in one band; F = D - (R - 1) and C=R For close packed detectors, scanned in B bands; F = BD - (R - 1) and C = BR For detectors spaced apart by S times the detector length and scanned in one band F = SD - S(R - 1) and C = SR A typical thermal imaging system might have 64 close-packed detectors scanned in 2 bands with an overscan factor of 8. In this case F = 121 and C = 16.

The scan generator of Fig. 1 can be modified to realize step-down scanning. For exam ple, for an overscan factor of 4 and one band, the mirror drum 5 is provided with 8 facets in two identical groups of four, each group comprising four adjacent facets. The angles that each of the facets in a group make with the pyramid axis 8 deviate, in succession, from 45 by amounts corresponding to the stepdown required from each facet. The deviation for a step-down of one detector spacing is given by half the ratio of the detector spacing to the focal length of the objective lens 2, allowing for reflection of the essentially parallel radiation by the facet. For example, for detectors close packed with 100 micron centre-to-centre spacing and an objective focal length of 50 millimetres, the angular subtense of adjacent detectors in 2 milliradians are hence the angular deviation at the facets for one step down is 1 mrad. or 0.057 . Diamet rically opposite facets on the mirror drum 5 have the same angular deviation for correct picture reconstruction. The scan generator of Fig. 1 can also be modified to provide the banded scanning cycle shown in Fig. 3. The mirror facet for scanning column 26 is set to deviate in angle from that of column 25 by eight times the deviation for a single detector step-down, i.e at 8 mrad. The deviations for the facets for columns 27, 28, etc to 32 are then 1 mrad, 9 mrad, 2 mrad, 10 mrad 3 mrad and 11 mrad respectively.

In a like manner, the pseudo-random scanning cycle referred to above can be realised.

For example, where an overscan factor of 4 is used, the deviations for the four succeeding facets of the group might be 0 mrad, 3 mrad, 1 mrad, and 2 mrad.

The invention has been described herein in relation to thermal imaging systems. However, it will be understood that the invention is not limited to any particular wavelength range, such as the infra-red range, for example.

Rather, the invention has wide applicability to systems in which arrays of detectors are utilised in such a manner that discrete signals are obtained from each detector.

Claims (5)

1. A thermal imaging system comprising an infra-red optical system for focussing a thermal image of a field of view upon a linear array of thermal radiation detectors, a scan generator for scanning said thermal image relative to the array in a repeated scanning cycle, and means for reconstructing a visible version of the field of view from signals supplied by the detectors, characterised in that the detectors are regularly pitched along the detector array and in that the scan generator comprises means for scanning the thermal image transversely relative to the length of the array in such a way that the scanning cycle comprises a group of parallel scans in which the thermal image is regularly displaced between adjacent scans along the array length by a multiple of the detector pitch.

2. A thermal imaging system as claimed in claim 1 wherein the scan generator is adapted to execute the scans of the group in a pseudo-random sequence such that each scan of the group is made once only and in the same relative order in each group, but that the order of the scans as a function of displacement of the thermal image along the length of the array within the group constitutes a random pattern.

3. A thermal imaging system as claimed in claims 1 or 2, in which the scan generator comprises a mirror drum having plane mirror facets set at evenly spaced angles around the drum and at angles relative to the drum axis to provide said regular displacements of the thermal image, when reflected by said facets relative to the array, and means for rotating the mirror drum about the drum axis to effect said repeated scanning cycle.

4. A thermal imaging system as claimed in any one of the preceding claims in which the scan generator comprises means for displacing the thermal image relative to the array along its length by an integral multiple of the length of the array between succeeding scans of the group, means for then performing a scan, and means for then restoring the thermal image to its position in the group relative to the array appropriate to the following scan of the group.

5. A thermal imaging system substantially as herebefore described with reference to Figs. 1 and 2, Figs. 1 and 3 or Figs 1 and 4 of the accompanying drawings.