GB2248742A - Optoelectronic system for analyzing video images - Google Patents

Description

OPTOELECTRONIC SYSTEM FOR ANALYZING VIDEO IMAGES OBTAINED BY SCANNING A DETECTOR BAR The present invention relates to an optoelectronic system for analyzing video images obtained by scanning a photodetector bar.

The invention is used mainly in cases where the scanning method adopted is of the circular type, but can also be suitable for uniaxial linear scanning transverse to the bar. It relates more specifically to the production of correction means for the purpose of eliminating image defects attributed to the individual variations in characteristics of the various detector elements of the Sar, the changes in continuous level and gain being corrected at each image and for each detector element.

It is envisaged that the invention will be used more particularly for missile homing systems with infra-red imaging.

In a video-frequency imaging system which uses image scanning by means of a bar possessing several photodetector elements, the image consists of an assembly of lines, each corresponding to a different detector of the bar.

These lines are concentric circles in the case of circular scanning and segments of parallel straight lines in the case of uniaxial linear scanning transverse to the bar.

Because in the infrared range, (particularly in the 8-12 micron band) the amplitude of the modulation constituting the useful part of the signal is only of the order of one thousandth of the total amplitude of the signal at the detector output, a very slight variation in the characteristics of the detector, especially its res ponse factor, causes a very substantial modification of the useful signal detected.

As regards a bar possessing several detectors, the variations in characteristics from one detector to the next, which are attributed, for example, to slight variations in the temperature of the bar, are usually different, thus resulting in the appearance of lining in the image, and this lining can damage or mask the useful content of this image.

The solution conventionally adopted for eliminating this lining involves eliminating the direct-current component of the signal at the output of each detector by means of filtering via a capacitor connected in series, for example, and restoring a relative direct-current component from line to line by means of a detector alignment device.

This alignment device is necessary since the detectors analyzing different zones of the image can generate, particularly in the case of contrasted images, signals the average modulation value of which is very different from line to line, thus resulting, again, in a lining effect on the image.

A method of alignment involves, for example, making the bar scan a structural part of uniform temperature (temperature reference) located at the edge of the field, measuring the corresponding signal obtained from each detector and adding or subtracting on each of the channels a direct-current component which is such that the corrected signals corresponding to the reference structure are all equal. A solution of this type is described in the patent publication FR-A-2,477,349.

Such a method of correction can be used easily for uniaxial linear scanning at the expense of an increase in optomechanical complexity, but it proves difficult to transfer it to circular scanning. Moreover, the disadvantage of this alignment solution is that the direct-current level is corrected for a reference temperature which is generally different from the average temperature of the scene, thus causing residual lining of the image attributed to the variations in gain between the detectors.

The object of the invention is to eliminate the image lining by making it possible to correct the variations in direct-current level and, if appropriate, the variations in gain from one detector to another.

The invention applies more particularly to analysis systems carrying out circular field scanning by means of a photodetector bar and especially to such systems mounted on missile homing devices with infrared imaging (or pseudo-imaging).

According to the invention, it is proposed to provide an optoelectronic system for analyzing video images obtained by scanning a bar of photodetector elenents, comprising receiving optics which produce the image of an observed field in a plane in which the bar is positioned, optomechanical means for cyclically executing a specific image scan in this plane and for allowing zone-by-zone analysis of the entire image by the said detector elements, means of processing the signals detected, equipped with alignment circuits for rectifying the lining of the image, the said system being defined in that the said optomechanical means are designed to ensure that, during each image scanning cycle, each of the said detector elements analyzes in succession two different zones of the image during two successive half-cycles, and in such a way that each of the said zones is observed successively, at least partially, by two separate detector elements during each cycle, by a first detector element during a first half-cycle and by a second detector element during the second half-cycle, the said processing means comprising circuits for aligning the continuous, which carry out a comparison of the signals detected by each pair of detectors analyzing one and the same zone, to prepare correction signals which equalize the average values of the signals detected by the detector elements.

The particular features and advantages of the present invention will emerge from the following description given by way of example, with the aid of the attached Figures in which: Figure 1 shows a diagram relating to the execution of a circular scan by a rotating optical device; Figure 2 shows the epicyclic path obtained for a circular scan with off-centering, according to the present invention; Figure 3 shows a diagram of a circular scan which is not off-center, using a bar positioned diametrically and non-symmetrically in relation to the center of rotation, to ensure that the entire image is covered; Figures 4 and 5 show diagrams of the reverse epicyclic paths obtained in the arrangement according to Figure 3, with, in addition, off-centering; Figure 6 shows a general diagram of an optoelectronic video-image analysis system according to the invent ion;; Figures 7 and 8 show exemplary embodiments of a system according to Figure 6 in gyro-stabilized versions which can be used particularly to constitute homing devices; Figure 9 shows a diagram of an embodiment of the processing circuits used for aligning the direct-current level of the detectors; Figure 10 shows a diagram of an embodiment of the processing circuits used for aligning the gain of the detectors; Figures 11 and 12 show partial diagrams relating to alternative embodiments of this system, using a linear or staggered radial bar; Figure 13 shows a diagram relating to the offset scan to be executed in front of the bar in the event of a Linear passage of the image transverse to the bar; Figure 14 shows a diagram relating to the switching circuit for the detection channels in the case of a uniaxial linear scan; ; Figure 15 shows a partial diagram of an embodiment of the optomechanical part for obtaining a uniaxial linear scan.

The principle put into effect according to the invention involves using the image itself as a temperature reference. For this purpose, the image scanning mode is modified so that each photodetector element of the bar analyzes at least two different zones of the image during each scanning cycle. Moreover, the arrangement is such that any two adjacent detector- elements each observe in turn one and the same image zone during one and the same particular analysis cycle. It thus becomes possible to compare the signals relating to this same image zone which are obtained from the two detector elerents and deduce from this the differences in the gain and direct-current level characteristics of the two detectors in question.

When this type of correction is used for all the detector elements two by two in succession (element of rank j with that of rank j+1, then the latter with that of rank j+2, etc.), the entire video image detected can be corrected.

The comparison of the signals obtained from the detectors of order j and j+1 in the overlap zone can be made in the following way: - the average value of each of the signals detected is measured, and the direct-current level of one of the detection channels is corrected in relation to the other, by determining a correction value which is such that the new average values are equal; - moreover, the root-mean-square value of the signals detected is measured, and a multiplying coeffi cient is determined so as to correct the gain defects.

It is also possible to carry out a gain correction by comparing the peak-to-peak value of the modulation of the signal detected in each channel, that is to say by each element.

This method is used more easily in optoelectronic systems where the analysis of the video image is carried out by means of circular scanning of the bar and is described a priori within such a framework.

The circular analysis of an image is obtained by rotating about its invariant axis an anisotropic optical device which is inverting along a certain longitudinal plane and non-inverting along the perpendicular longitudinal plane, the invariant axis being the line of intersection between these two planes.

Figure 1 recalls the characteristics which are necessary and sufficient for executing a circular scan by means of a rotating optical device. In a longitudinal sectional plane, called a non-inverting plane PNI, the device forms a direct image of an object; thus, the object CB has the image C'B' in the same direction. In a second longitudinal sectional plane, called an inverting plane PI and perpendicular to the preceding plane, the optical device forms an inverted image of an object; thus, the object CA will have the image C'A' in the opposite direction. Magnification has not been taken into account in this diagram. It appears that the device is equivalent to a symmetry in relation to a straight line in the image plane.If the optical device D is rotated through an angle e, the image rotates in the same direction through a double angle 2e , this property being general for all these types of device.

To obtain a perfectly circular image, the axis of rotation (at the speed ) of the optical device must be either merged with or parallel to its invariant axis (in the case of a system of the afocal type); in this case, the instantaneous center of rotation of the image is fixed at a point 0 corresponding to the line of the mechanical axis of rotation in the image plane. If this condition is not satisfied, that is to say the axis of rotation of the optical system is not parallel to that of the invariant axis, the rotational movement of the image (at the speed 2X) is modified by a superimposed movement of circular displacement (at a speed cn in the same direction) of the instantaneous center of rotation 0' along a circle centered on the abovementioned point 0.

The path of a point of the image (or that brought into the object plane from a point on the bar) is then no longer a circle with center 0, but an epicycloid with instantaneous center 0', as illustrated in Figure 2.

A simple geometrical calculation shows that this path can be assimilated to a good approximation (if Ro > 2e) to a curve formed from four portions of circles: - a first circle of center C and radius R (from a to b) - a second circle of center A' and radius Ro (from b to c) - a third circle of center C' and radius R' (from c to d) - a fourth circle of center A and radius Ro (from d to a).

With Ro being the radius of the point M rotating at the speed 2 in relation to the instantaneous center of rotation 0', and with e being the off-centering, namely the radius of the circle centered on 0 and on which the point 0' moves at the rotational speed w,

FxC = -xC' = e (1 - 1//2) lvC = yC' = o {yXAA : xA' = o yA = -yA' = e - 1) R = Ro + el ff R' = Ro - e/4 kx = 2e(4- 1) The point M considered to be in the image plane will travel along the epicycloid during an image scanning period. This period or cycle can be subdivided into two half-cycles.

During the first half-cycle, the center 0' is displaced 1800 on the circle of radius e centered on 0, for example from 0'1 to 0'2 in the direction indicated, taking 0'1 as the original position, whilst the point M travels 3600, going from M1, the position considered to be the original position, up to M2, passing via the points b and c. During the second half-cycle, the point 0' travels along the remaining half-circle from û'2 to 0'1, and the point M executes a new revolution, going from M2 to M1, this time passing via the points d and a in succession.

When, according to the invention, the image is analyzed by means of a bar arranged diametrically in relation to the center of rotation of the image, the paths of the even points (that is to say, those corresponding to the detectors on one side of the center) and odd points (the detectors on the other side of the center) are inverted epicycloids (offset a half-revolution), thus making it possible, by means of a careful choice of parameters, to make the circles representing the paths of the consecutive points coincide with one another two by two.

This is shown with the aid of the following Figures 3 and 4.

Figure 3 shows a bar arranged diametrically with the detectors bearing odd numbers D1, D3, etc. being towards the top and those bearing even numbers D2, D4, etc. being towards the bottom. To ensure overlapping, the bar is not arranged symmetrically in relation to the point 0 corresponding to the line of the axis of rotation, but is offset as a result of translation along the axis x, so that each pair of detectors D, Dj+l or Dj, Dj-l has a common analysis zone.

The offset denoted by ri is measured from the center of the detector D1 nearest to O at this point.

Its value is advantageously r1=P/2. A circular scan of center 0 without off-centering has been considered in this representation. The parameter P represents the pitch of the image corresponding to half the pitch of the bar (the distance 2P between the centers of two successive detectors). The hatched circular zone represents the overlap zone located between two successive detectors, for example D2 and D3.

Figure 4 corresponds to a diametral arrangement of the detector bar, this time with an off-center circular scan.

The value of the off-centering 00' of the axis of rotation 0, is e=PT722 in relation to the invariant axis at 0 and the offset rl=P/2. The paths represented correspond to the detectors D3 and D4 symbolized by their respective centers. The circles of radii R3 and R'4 are shown adjacent to one another. They coincide exactly when the values of the parameters e and rl are those mentioned above.

For the sake of argument, it is equivalent to consider the image fixed and the bar driven in rotation with off-centering in order to execute the scanning of the image. Furthermore, as a first approximation and for the sake of simplification, the epicycloid is considered to take the form of the circles R and R' during two successive frames of a scanning cycle.

Thus, during a complete scanning cycle, the detector D3 describes a circle of radius R3 during the first frame, whilst the detector D4 describes a circle of radius R'4 equal to R3, then, during the second frame, a circle of radius R'3 less than R3, whilst the detector D4 describes a circle of radius R4 greater than R'3.

According to this reasoning, the detectors are considered to be point detectors reduced to their center.

Consequently, during a first half of the cycle or first frame, there is a coincidence between the paths of the detectors 3 and 4 on the image, thus making it possible to utilize the signals obtained from these two detectors during this half of the cycle in order to compute the corrections to be applied.

When the situation of all the detectors of the bar is considered, the two different frames of a complete analysis cycle can be distinguished: the even frames (first half-cycle) where there is coincidence between the detectors number 1 and 2, 3 and 4, 5 and 6, etc., and the odd frames (second half-cycle) where there is coincidence between the detectors number 2 and 3, 4 and 5, 6 and 7, etc. Figure 5 illustrates these respective coincidences of paths for the values of the parameters e and rl in question. At the end of the complete cycle, it has been possible to compare all the detectors of the bar two by two.With different values of the paraneters, the coincidence of the paths of the detectors in the comparison zone can be improved; thus, in the example considered in Figure 5, the off-centering can be adjusted and can be given a value slightly greater than P/7 so as to move in this direction.

The invention can be used for any type of circular scan. For example, the arrangement illustrated diagrammatically and in simplified form in Figure 6 may be considered, this incorporating receiving optics 1-2 which produce the image of the field observed in a plane in which a photodetector bar 3 is located. The receiving optics consist of a Cassegrain arrangement with a main mirror 1 which reflects the radiation towards a secondary mirror 2 consisting of a reflecting dihedron. The axis 2 represents the optical axis or invariant axis of the arrangement which focuses the radiation on the bar 3 arranged diametrically in the focal plane.

A right-angled dihedron consisting of two plane mirrors placed in an optical path supplies an image symmetrical in relation to the line of intersection of the two faces of the dihedron, called the edge of the di he- dron. This geometrical optical property is well known, and if an axis of rctation perpendicular to the edge is considered, there is an inverting plane containing the axis of rotation and perpendicular to the edge of the dihedron and a non-inverting plane containing the axis of rotation and the edge of the dihedron. The arrangement with a rotating right-angled dihedron thus solves the problem of the circular scanning of the field.

In the Cassegrain arrangement under consideration, the dihedron or the optical system as a whole 1 and 2, as shown, is driven in rotation; this solution is parti cularly suitable for integration in a homing device with imaging. The block 4 symbolizes the means of driving in rotation, and these means can consist of the head of a gyroscope in a gyro-stabilized version. According to the invention, the axis of rotation Zo is inclined relative to the invariant axis Z at an angle a, thus giving rise to the epicyclic path described above; the rotating optical system 1-2 is adjusted so that the angular variation between the axis of rotation Zo and the invariant axis Z of the optical system is equal to e/2F radians (e = off-centering brought into the image plane and F = the focal length of the optical system).The other condition which is satisfied is that the detector bar 3 is positioned diametrically in relation to the point 0' of the image plane pierced by the axis of rotation Zo of the movable unit, the distance rl between the center cf te first detector D1 and the instantaneous center of rotation 0' being maintained.

Angular detection circuits 6 are provided to generate a synchronizing signal ST which subsequently makes it possible, during processing, to distinguish the successive even and odd frames during the image period T corresponding to one revolution. This can be achieved by means of an optical track and an assembly of photoemitting and photoreceiving diodes. The signal ST comprises a first pulse of duration T/2, corresponding to a first half-cycle, and a second pulse of opposite sign, representing the duration T/2 of the second half-cycle.

In gyro-stabilized solutions, such as the examples illustrated in Figures 7 and 8, the optical system 1-2 is driven in rotation by the gyroscope head, and a simple means of carrying out the adjustment of the angular variation a is to modify the dynamic balancing of the gyroscope head. Screws in tapped receptacles are usually provided to adjust the exact coincidence between the axes Z and Zo in the case of habitual use; action on these elements results, in the same way, in the desired offset a according to the method used in the present invention. In practice, this offset remains of low value, for example of the order of one milli-radian.

The video signals detected are preamplified before being transmitted to an assembly of processing circuits 5 for equalizing the average values of the directcurrent level and, if appropriate, for compensating the differences in gain from one detector to another. For this purpose, the assembly 5 possesses a switching circuit 50 consisting of an analog switch controlled by the synchronizing signal ST, this switch therefore being controlled in synchronism with the rotation in order to execute the correct pairings of the signals detected, at the same time as the above-described coincidences during the successive frames. The circuit 51 downstream of the switch equalizes the average direct-current levels.If there is also the intention to act on the respective gains, a second circuit 52 is located upstream of the switch 50. In each of these circuits, for each channel resulting from a pairing of detectors, a circuit, such as the circuit 12 or 24, prepares a compensating signaL.

In the case of compensation of the direct-current Level, this signal SCN is summed in a summing circuit, such as 15, with the signal detected. In the case of gain compensation, the corresponding signal SCG controls the gain of a corresponding amplifier, such as 22. These circuits will be analyzed in detail subsequently.

Reference will now be made again to Figure 7 which illustrates a method of using the image analysis system on a homing device with infrared imaging. The optomechanical scanning device consists of the Cassegrain receiving optics 1-2 with dihedron, driven by the gyroscope head 30. Downstream of the focal plane, image transposition optics make it possible to shift the bar to the center of the gyroscope and make it integral with the body of the missile. These additional optics consist of a diaphragm 31 which delimits the observed field, a first lens 32 which, together with the Cassegrain arrangement 1-2, constitutes an afocal optical system, and a second lens 33 which picks up the parallel beam coming from 32 in order to focus it on the bar 3. The latter is located inside a cooling device 34, for example a cryostat.The assembly is shown in the conventional position, that is to say aligned with the longitudinal axis of the missile. The optical elements 1-2-31-32 are driven in rotation by the gyroscope head or rotor 30.

The components of the rotating part are represented by means of dotted lines instead of hatching, and those of the fixed part have been left clear, in order the better to distinguish these parts and simplify the representation. The ballbearings 35 make it possible to uncouple mechanically the terminal optical system 33 from the rotation of the gyroscope head. The latter incorporates an annular magnet 36 which undergoes the magnetic effects of the precession coitus 37 mounted fixedly on a mechanical support integral with the body 38. The upper part represents a longitudinal section passing through the azimuth axis YG of the gimbal arrangement, and the lower part corresponds to a longitudinal section, at 900 relative to the preceding section, passing through the elevation axis XS, to give a better visual idea of the gimbal suspension.Bearings 41 and 42 allow respective rotations in azimuth and in elevation. The mirror 2 is supported by means of arms 43 of small thickness which connect it mechanically to the gyroscope head. The diaphragm 31 makes it possible to limit the entrance of in terfering radiation. Moreover, the lenses 32 and 33 possess a central deposit corresponding to the zone not used for reception by this Cassegrain arrangement.

Figure 8, in a partial diagram, illustrates another embodiment without additional optics, with the detector 3 arranged directly in the focal plane of the Cassegrain receiving optics 1-2. The detector is uncoupled from the rotation by means of the ballbearings 35.

Each image signal detected SOj consists of a direct-current component corresponding to the average value and attributed essentially to noise and of a variable component (of zero average value) which essentially represents the useful signal. The non-uniformity of sensitivity from one detector to another results in different responses in respect of one and the same continuous background analyzed, and this defect must be rectified, this being effected by means of a circuit 51 which equalizes the continuous levels of the various detectors of the bar.In contrast, different responses in respect of one and the same variable useful component result from differences in the gain of the detector ele events; defects of this order are less pronounced, and it is usually not indispensable to instal a correction circuit 52 provided for this purpose.

Figure 9 shows a block diagram of the circuit 51 for correcting the continuous level (called offset according to an English designation). The three detector channels of rank 1, 2 and 3 have been considered, and the following ones are deduced in the same way; the circuit relates to analog processing in the case of a bar 3 with discreet elements. This functional diagram can easily be transferred to other particular cases, such as that of a detector with multiplexed elements, or to the case of the digital processing of signals.

After amplification, the signal from each detector is cut off by the analog switch 50 synchronous with the rotation of the gyroscope head.

Figure 9 shows it in the position A corresponding to the even frame, B being the position during the following odd frame, and so on and so forth. The circuit is designed so that the divergencies of continuous levels of each detector are corrected and the average level of the image remains constant, equal to a reference value VR corresponding to a predetermined image temperature.

The circuits 11, 12, 13 (Figure 7) for preparing the compensating signals are formed with an input storage integrator, such as 60 and 61 respectively for the signals SD1 and SDt during the even frame in question. The inputs SD1 and SD2 can be written in the form SD1 = 01 + S12 and SD2 = 32 + S21, in which 01 and 02 are the continuous components (or offsets) integrated at 50 and 51, and S12 and S21 are the useful components of zero average value and moreover which, for this pair of detectors, correspond to one and the same image zone during the even frame; consequently, the values SI 2 and S21 are equal, and the outputs of the integrators represent in practice (apart from the gain errors) the continuous values 01 and 02. The subtracting circuit 62 which follows provides the difference 01-02 between the integrated signals, and therefore the offset variation between the detectors D1 and D2. When the offset variation of the detector D1 in relation to a reference value OM, namely OM-01, is added in a summing circuit 63, the latter supplies the offset variation of the detector D2 in relation to OM, namely OM-02. This value constitutes the offset compensation signal to be applied to the signal SD2 in order to align it with the reference value OM, this latter operation being carried out in the summing circuit IS.

The circuit 11 which gives the alignment reference is designed differently. It incorporates the integrator 64 which receives the signal SD1 during the odd frame (the switch 50 in position B) and a subtracting circuit 65 in which the offset component of D1 is subtracted from a reference value. This offset component has been indicated as 01 for the sake of simplification, but its value can differ from this because the analysis during the odd frame does not correspond to the same image zone for this detector as during the even frame.

On the other hand, it must be considered that the continuous background analyzed can have considerable variations both as regards the image as a whole and at the level of a detector (a bright spot). To remedy all this, the organization of the alignment reference takes into account the offset variation of the image as a whole in relation to a reference VR.This is obtained in the circuit 53 which comprises in succession an input summer 66 for summing all the signals SD1 to SDn detected, a storage integrator 67 which integrates the sum of the signals, a divide-by-n circuit 68 (n being the number of detector elements), to bring the integrated value to the average offset value presented by the image, and a differential amplifier 69 for obtaining the signal proportional to the difference between this average continuous image value and the selected reference value VR. The output OM represents the offset deviation of the image in relation to VR and makes it possible by means of the summer 14 to correct the deviation of the detector D1.

It is subsequently applied to the circuit 12 to correct the deviation of the detector D2 by means of the circuit 15, and so on and so forth, in order to correct the detector Dj+1 in relation to the detector Dj. It will be noted that the value OM is shifted by degrees; if it is considered that the output of the integrator 64 supplies a value 0'1 different from 01, so that 0t1 = 01+ dO1, this variation dOl will be added to the component OM in terms of its value and sign, and the alignment of the continuous level desired for the detectors is obtained in any case.It will also be noted that there is a filtering effect produced by the integrator circuits 60, 61, 64, etc. in relation to abrupt variations in the signal detected, for example as regards localized bright spots. The channel of the detector D1 is selected to constitute the reference, since this detector is nearest to the center of rotation 0', and the corresponding analysis zone is the smallest of the image.

The gain correction circuit 52 is indicated diagrammatically in Figure 10. In this circuit, a comparison is made, for each pair of detectors, of the peak-topeak value of the variable component corresponding to the modulation, during the frame corresponding to the coincidence zone, and this comparison makes it possible to correct the gain errors by degrees.

This circuit is preferably located upstream of the continuous level correction circuit 51, since any difference in gain results in a difference in continuous level, over a limited zone, in a signal of which the average value over this zone is non-zero. It is therefore necessary for the gains to be corrected before the continuous levels are corrected. The time constant of the variable-gain amplifiers can be relatively high, the variations in gain usually being somewhat slow.

The diagram of Figure 10 corresponds to an exem plary embodiment. Each circuit generating a gain compensation signal SCGj possesses for each detection channel positive-peak detection circuits 70, 72 and negative peak detection circuits 71, 73, with storage and resetting to z ro at the image rate T. The peak values detected are s btracted in an associated circuit 74, 75, to obtain the amplitude of the modulation of the correspond.

ing channel SDj during the coincidence frame in question. A terminal differential amplifier circuit 76 supplies the compensating signal proportional to the difference between the outputs of the subtracters 76 and 75. This signal, such as SCG2, is applied to the variable-gain amplifier 22 of the channel D2 in order to align the gain with that of the channel D1. The latter need have only one ordinary amplifier 21 with a gain equal to the average gain of the other amplifiers, such as 22.

Another possible solution is to compare the root-mean-square values of the modulation instead of the peak-to-peak values, thus making the circuit less sensitive to noise, but the circuits used are more complex.

In the event that the signal is multiplexed and digitized, the same types of processing can be carried out in wired digital form or by software with a sufficiently powerful and high-speed computation unit.

In general, the correction of gain is not necessary since the corresponding variations are slight, and only the correction of the continuous level must be provided. It should also be noted that the chains amplifying the signals coming from the bar can have continuous connection or capacity of connection, and this can reduce the dynamics of the signals to be processed.

The invention has been described above in terms of the use of a rectilinear detector bar mounted diametrically, and in terms of a circular scanning mode with off-centering. Other alternative forms are possible using detectors with a bar of a different type.

Figure 11 shows another configuration of a detector using a radial rectilinear bar; again, there are coincidence zones from one detector to another, but over half an image revolution only. Moreover, to prevent blind zones, it is necessary for the detectors to be very close to one another; for example, in the case of a pitch of the order of 100 microns, the inter-detector space ought to b approximately 10 microns, and this presents difficulties in terms of execution.

The other example of a configuration, shown in figure 12, represents the case of a radially staggered bar. This type of component is more readily available.

The scanning pattern is a little more complex, but there is, again, a coincidence zone over approximately half a revolution. In addition to a smaller overlap zone presented by these two alternative forms, the image analysis is also more complex than in the case of a diametral bar.

Consideration will now be given to the case of sytems which use uniaxial linear scanning in a direction perpendicular to that of the detector bar. The linear passage of the image is repeated periodically and prefe ratty in alternation. A complete cycle thus comprises (Figure 13) an outward horizontal scan, for example from left to right, the bar being considered to be vertical, and a return horizontal scan from right to Left.

According to the characteristics dealt with above, the system must be designed, in this case, so that the relative displacement between the image and bar corresponds to the offset between two successive detectors, that is to say, to the pitch of the bar. Thus, the superimposition of the paths of the even and odd detectors (the zones analyzed by these detectors) alternately on the image is obtained. This alternating vertical offset is obtained by preferably acting on the optomechanical scanning means which execute the uniaxial linear scan, rather than on the detector, this being for various reasons, particularly convenience. There is also an offset cf one half-cycle between the passage of two consecutive detectors over the same image zone.

During the outward scan corresponding, for example, to an even frame, the signals of the detectors of rank j are compared with those of rank j-l of the preceding return scan (D3n compared with D2(n-1)) which will have been memorized. During the following return scan corresponding to the odd frame, the signals of rank j are compared with those of rank j+l of the preceding outward scan (D3(n+i) compared with D4n) which will have been memorized. The diagrams of the correction circuits are the same as those described for circular scanning, but with a difference as regards the frame synchronizing switch 50 as shown in Figure 14.

As regards the optomechanical execution of the scan with frame translation, the problem is much less simple th n in circular scanning. An exemplary embodiment is given in Figure 15. The principle adopted involves oscillating the secondary mirror 82 of a Cassegrain arrangement 81-82 through a small angle to obtain the desired vertical offset. Horizontal passage is obtained for it as a result of the periodic oscillation of a blade 83 about a vertical axis R. The movement of the mirror 82 about a horizontal axis R2 perpendicular to the draving plane is produced at the end of each travel of the blade 83, thus resulting in the desired scan indicated in Figure 13. The vertical offset is produced by the use of piezoelectric jacks 84 and 85 supplied in opposition, so that one contracts while the other expands, thus avoiding the need to introduce image defocussing.

Claims (15)

1. A optoelectronic system for analyzing video images obtained by scanning a bar (3) of photodetector elements, comprising receiving optics (1-2) which produce the image of an observed field in a plane in which the bar is positioned, optomechanical means (2-4) for cyclically producing a specific image scan in this plane and allowing one-by-zone analysis of the entire image by the said detector elements, means (5) of processing the signals detected, equipped with alignment circuits for rectifying the lining of the image, characterized in that the said optomechanical means are designed to ensure that, during each image scanning cycle, each of the said detector elements analyzes in succession two different zones of the image during two successive half cycles, and in such a way that each of the said zones is observed successively, at Least partially, by two separate detector elements during each cycle, by a first detector element during a first half-cycle and by a second detector element during the second half-cycle, the said processing means comprising circuits for aligning the continuous level (50, 51), which compare the signals detected by each pair of detectors analyzing one and the same zone, in order to prepare correction signals which equalize the average values of the signals detected by the detector elements.

2. A system as claimed in claim 1, characterized in that the processing circuits also incorporate gain alignment circuits (52) which prepare correction signals making it possible to equalize the gains of the photodetector elements.

3. A system as claimed in claim 1 or 2, in which the optomechanical scanning means use a rotating optical device to execute a circular scan about the invariant axis, characterized in that the axis of rotation of the said optical device is inclined (a) in relation to the direction of the invariant axis, so that the path of the scan becomes epicyclic.

4. A system as claimed in claim 3, characterized in that the bar (3) is positioned diametrically with a specific offset (r1) in relation to the axis of rotation, which has an off-centering (e) in relation to the invariant axis resulting from the said inclination.

5. A system as claimed in claim 4, characterized in that the bar is arranged diametrically on either side of the line (O) of the invariant axis, the elements of odd rank being considered on one radius and those of even rank on the opposite radius, each of the said pairs of detectors comprising an even detector and an odd detector of successive rank.

6. A system as claimed in claim 4 or 5, characterized in that the said offset (tri) and the said off-centering (e) are determined in such a way that any one of the epicyclic paths relating to a first detector element coincides, at least partially, during a first-half cycle with a second epicyclic path relating to a second detector element and, during the second half-cycle, with a third epicyclic path relating to a third detector element, the said second, first and third detector elements being of successive rank.

7. A system as claimed in any one of the preceding claims, characterized in that the receiving optics consist of a Cassegrain arrangement with a main mirror (1) and a secondary mirror (2) consisting of a reflecting right-angled dihedron, the latter or the Cassegrain assembly being driven in rotation to execute the said offcenter circular scan.

8. A system as claimed in any one of the preceding claims, characterized in that the receiving optics are gyro-stabilized.

9. A system as claimed in any one of the preceding claims, characterized in that the continuous level align ment circuits incorporate an analog switch (50) synchronous with the scan, for switching the pairs of detector elements during their analysis of one and the same zone, this being followed by an actual alignment circuit (51) comprising a plurality of comparator circuits, a first comparator circuit (11) constituting the alignment reference, taking into account the variation in continuous level of the image as a whole in relation to a predetermined value (VR).

10. A system as claimed in claims 2 and 9 taken as a whole, characterized in that the gain alignment circuit (52) comprises variable-gain amplifiers interposed in the detection channels upstream of the said analog switch, and comparison means forming a gain control loop for each amplifier, the input of the loops being switched by the analog switch.

11. A system as claimed in claim 1 or 2, characterized in that the optomechanical scanning means execute a uniaxial linear scan transverse to the bar, with an offset equal to the pitch of the bar from one half-cycle to the next, so that the detector elements of order j and j+l analyze the same zone, the first during a first halfcycLe and the second during the following half-cycle.

12. A system as claimed in claim 11, characterized in that the optomechanical means and the receiving means consist together of a Cassegrain arrangement with a main mirror (81) and a secondary mirror (82) for focussing the radiation, and an oscillating mirror (83) for producking the periodic transverse passage, the said offset being produced by piezoelectric jacks (84-85) which control the positioning of the secondary mirror.

13. The use of a system as claimed in any one of the preceding claims for constituting a missile homing device in which the head of a gyroscope drives Cassegrain receiving optics.

14. An optoelectronic system substantially as hereinbefore described with reference to, and as illustrated in, the accomp2nying draWings.

AMENDSMENTS TO THE CLAIMS HAVE BEEN FILED AS FOLLOWS CLAIMS 1. An optoelectronic system for processing and analyzing video images comprising a bar of photodetector elements, receiving optics which produce an image of an observed field in a plane in which the bar is positioned, optomechanica means fo cyclically producing a specific image sca:: this plane and allowing analysis of the entire image in zones observed by said detector elements, means for processing the signals detected and which are equipped with alignment circuits operable to avoid a lining effect of the image, wherein said optomechanical means are designed to ensure that, during each image scanning cycle, each of said detector elements analyzes tlfO different zones of the image one after the other, one per half cycle, and in such a way that each of said tlfO zones considered by one of the detector elements is observed, at least partially, by another of the detector elements during each cycle, the other detector element that considers one of the two zones being different from the other detector element that considers the other of the tao zones, said processing means comprising continuous level alignment circuits in which the signals detected by each pair of detector elements that analyze the same zone during a cycle are compared in order to generate correction signals to equalize the average values of the signals detected by the detector element of said bar.

2. A system as claimed in claim 1, wherein the processing means also incorporate gain alignment circuits which prepare correction signals making it possible to equalize the gains of the photodetector elements.

A system as claimed in claim 1 or claim the optomechanical scanning means use S rotating optic device to execute a c-ircular scan about the invariant axis, wherein the axis of rotation of said optical device is inclined in relation to the direction of the invariant axis so that the path of the scan becomes epicyclic.

4. A system as claimed in claim 3, wherein the bar is positioned diametrically with a specific offset in relation to the axis of rotation which is off-center in relation to the invariant axis as a result of said inclination.

5. A system as claimed in claim 4, wherein the bar is arranged diametrically on either side of the invariant axis, the elements of odd rank being considered on one radius and those of even rank on the opposite radius, each of said pairs of detectors comprising an even detector and an odd detector of successive rank.

6. A system as claimed in claim 4 or claim 5, wherein said offset and said off-centering are determined in such a way that any one of the epicyclic paths relating to a first detector element coincides, at least partially, during a first-half cycle with a second epicyclic path relating to a second detector element and, during the second half-cycle, with a third epicyclic path relating to a third detector element, the said second, first and third detector elements being of successive rank.

7. A system as claimed in any one of the preceding claimes, wherein the receiving optics consist of a Cassegraim arrangement with a main mirror and a secondary mirror consisting of a reflecting right-angled dihedron, the latter or the Cassegraim assembly being driven in rotation to execute said off-center circular scan.

8. A system as claimed in any one of the preceding claims, wherein the receiving optics are gyro-stabilized.

9. A system as claimed in any one of the preceding claims, wherein the continous level alignment circuits incorporate an analog switch, synchronous with the scan, for stitching the pairs of detector elements that analyze the same zone during a cycle during their analysis of that zone, this being followed by an actual alignment circuit comprising a plurality of comparator circuits, a first comparator circuit constituting the alignment reference, taking into account the variation in continuous level of the image as a whole in relation to a predetermined value.

10. A system as claimed in claim 9 when appended to claim 2 or when appended to claims 2 and 3, wherein the gain alignment circuit comprises variable-gain amplifiers interposed in the detection channels upstream of the said analog switch, and comparison means forming a gain control loop for each amplifier, the input of the loops being switched by the analog switch.

11. A system as claimed in claim I or claim 2, wherein each of said two zones is analyzed partially and each of them is analyzed successively by a first detecting element of the corresponding pair of detecting elements during a first half-cycle and by the second detecting element 0 said pair during the second half-cycle.

12. A system as claimed in claim 1, claim 2 or claim 11, whereim the optomechanical scanning means execute a uniaxial linear scan transverse to the bar, with an offset equal to the pithc of the bar from one half-cycle to the next, so that two of the detector elements analyze the same zone, the first during a first half-cycle and the second during the following half-cycle.

13. A system as claimed in claim 12, wherein the opto- mechanical means and the receiving means consist together of a Cassegrain arrangement with a main mirror and a secondary mirror for focussing the radiation, and an oscillating mirror for producing the periodic transverse passage, the said offset being produced by piezoelectric jacks which control the positioning of the secondary mirror.

14. The use of a system as claimed in any one of the preceding claims for constituting a missile homing device in which the head of a gyroscope drives Cassegrain receiving optics.

15. An optoelectronic system substantially as hereinbefore described with reference to the accompanying drawings and as illustrated in Figures 1 to 10, or Figure 11, or Figure 12, or Figures 13 and 14, or Figures 13 to 15 of those drawings.