Detection device for a missile

Abstract

Detection device (1) for a missile, in particular a guided missile, comprising an optical device (2) configured to image radiation (5, 6) onto a detector plane (3) of a detector device (4) of the detection device (1) into at least one spot (7, 8), wherein the detection device (1) has a phase element (9) configured to generate a blur of the at least one spot (7, 8) in the detector plane (3), wherein a control device (10) is configured to transform the at least one blurred spot (7, 8) imaged in the detector plane (3) and having a first image sharpness into at least one second image sharpness with greater image sharpness than the first image sharpness by means of a transformation algorithm, in particular based on artificial intelligence.

Description

[0001] The invention relates to a detection device for a missile, in particular a guided missile, comprising an optical device configured to image radiation onto a detector plane of a detector device of the detection device in at least one spot, wherein the detection device has a phase element configured to generate a blurring of the at least one spot in the detector plane.

[0002] Detection devices for missiles, which include an optical unit by which radiation is focused into at least one spot or as a spot in the detector plane, are generally known from the prior art, for example for object detection or object tracking. For example, the missile can be controlled based on signals from the detector unit or the detection device.

[0003] It is further known from the prior art that the detection device can have at least one phase element with which a deliberate blur, for example a distortion or defocus, is introduced into the imaging of the radiation or into the spot imaged in the detector plane, so that it is distorted compared to an ideal image. This can be used, for example, to extend the depth of field. In this case, it is accepted that the imaging of the spot in the detector plane is impaired, but the image undergoes a comparatively smaller change over a longer area along the optical axis, particularly compared to an optimized sharp image of the spot in the detector plane.

[0004] Typically, processing the spot at the detector level requires that the change introduced by the phase element is precisely known. This means that the "prior knowledge" about the phase element must be exact, and further influences that may arise during the operation of the detection device cannot be taken into account.

[0005] The invention is based on the objective of providing an improved detection device for a missile.

[0006] The problem is solved by a detection device having the features of claim 1. Advantageous embodiments are the subject of the dependent claims.

[0007] As described above, the invention relates to a detection device for a missile, specifically for a guided missile. The detection device can detect radiation emitted, for example, by an object in the vicinity of the missile. Based on the detected radiation, the missile, for example in a guided missile configuration, can be controlled. In other words, the missile can be equipped with the described detection device for object detection and / or object tracking.

[0008] To detect the radiation, the detection device includes an optical unit. The optical unit comprises at least one optical element and / or at least one mirror. The optical unit allows the radiation to be introduced into the detection device from outside, or directed precisely into the detector plane of the detector device. Specifically, the radiation is imaged as a spot in the detector plane. Within the detector plane, the detector device thus has various detection areas, for example, "pixels," with which the radiation, which can be imaged as a spot onto the detector device, can be detected.

[0009] The detection device further comprises a phase element, which can be arranged in the beam path of the optical unit. The phase element introduces a blurring of at least one spot in the detector plane. In other words, a deliberate blurring is introduced into the image, so that the spot, compared to using the optical unit without the phase element, is not optimally focused, but rather defocused or distorted, i.e., generally "blurred," in the detector plane. As described, while the deliberate blurring reduces the sharpness of the spot in the detector plane, it increases or stabilizes the spot's movement over a comparatively large area along the optical axis. Therefore, the imaging of the radiation into the detector plane becomes more robust.

[0010] The invention is based on the finding that a control unit of the detection device is configured to transform, by means of a transformation algorithm, in particular based on artificial intelligence, the at least one spot, which is projected blurred into the detector plane and has a first image sharpness, into at least one second spot with greater image sharpness than the first. In other words, the spot is projected by the optical device together with the phase element with a first image sharpness in the detector plane. From this, a second spot can be transformed by the transformation algorithm, which is sharper than the first spot and thus has a higher image sharpness. Specifically, the blurring caused by the phase element can be reduced or eliminated by the transformation algorithm.

[0011] In one embodiment of the detection device, the control unit can be configured, by means of a transformation algorithm, to at least partially compensate for the blurring caused by the phase element. The transformation algorithm can be specifically trained on the effect that the phase element causes in or in conjunction with the optical device. The effect created by the introduced blurring, intended to increase the depth of field of the optical image, can thus be at least partially compensated for by the transformation algorithm. Advantageously, this can be implemented more robustly with respect to the image created in the detection device, whereby the transformation algorithm, after training, can also take further effects into account and, if necessary, requires less prior knowledge about the phase element.Deviations in the optical system, for example due to manufacturing tolerances, assembly tolerances, changes during operation, and the like, as well as corresponding effects of the phase element, can therefore be taken into account by the transformation algorithm. In other words, the transformation algorithm is designed to consider or at least partially compensate for at least one error in the optical system and / or the phase element when transforming the first spot into the second spot.

[0012] Accordingly, the blur introduced by the phase element can be selectively removed by the transformation algorithm. This allows the advantages of a greater depth of field to be combined with the benefit of a sharp image or a sharp spot for further processing. In particular, manufacturing tolerances, such as those related to the installation position of the detector, as well as changes to the detection device during operation, such as thermal changes, can be better compensated for, or the detection device can be made more robust against such influences.

[0013] In one embodiment of the detection device, the phase element can be used to create a stability zone along the optical axis of the detector. As described, the phase element reduces the maximum achievable sharpness or image sharpness in the imaging of the radiation as a spot on the detector, but increases the depth of field. This allows a stability zone to be created along the optical axis in which the imaging of the spot changes only slightly.

[0014] Advantageously, it is therefore not necessary to use a compensated optical system, which is comparatively complex and expensive. Temperature fluctuations, such as those occurring during missile operation due to temperature increases, particularly those resulting from cooling of the detector system and temperature rises during flight, can be compensated for, or the detection device is more robust due to the stability range. Likewise, the detection device can be more robust against manufacturing tolerances due to the generated stability range, for example, deviations of the detector system or the detector plane from a target position, especially along the optical axis.

[0015] The detection device may further be configured to include a control unit capable of transforming at least one spot based on phase information from the phase element, using a transformation algorithm. If the change generated by the phase element, or the change in the spot within the detector plane resulting from the phase element, is known, this can be used as phase information. This prior knowledge about how the deformation or change of the spot results from the use of the phase element can be fed into the transformation algorithm.

[0016] The transformation algorithm can therefore calculate the change in the spot due to the phase element and thus transform it into a second spot that would correspond to an image of the radiation onto the detector plane without the use of the phase element. In principle, any element can be used as a phase element, particularly one with an asymmetrical configuration. The geometry of the phase element can therefore have certain properties that modify the spot compared to an ideal image. For example, a cubic phase plate can be used as a phase element.

[0017] In one embodiment of the detection device, the phase element can be configured as a separate phase plate and / or as a component of an optical element within the optical system. Multiple phase elements can be used, or the phase element can be distributed across several elements within the optical system. This allows for a degree of flexibility in how the phase element can be arranged or configured within the beam path of the detection device to generate the blurring effect. For example, the phase element can be configured as a separate element, i.e., a phase plate, positioned within the beam path of the optical system. Furthermore, the surfaces of optical elements, such as lenses, mirrors, or the like, can be specifically modified to function as phase elements or to be configured as such.

[0018] In another embodiment of the detection device, the phase element can have an asymmetrical geometry, in particular a freeform surface. Specifically, the phase element is not rotationally symmetrical but has a specific geometry. As the radiation passes through the phase element, it is altered, resulting in a blurred spot imaged in the detector plane. This creates, as described, a stable region along the optical axis, making the detection device more robust against changes. Along the optical axis, neighboring planes can form in which the spot's appearance is very similar or changes only slightly. As described, the change introduced by the phase element can be factored out by the transformation algorithm, for example, as a deconvolution.

[0019] The detection device may also be provided that the detector unit has two groups of detection areas which are sensitive to different wavelengths and / or wavelength ranges of the radiation, wherein the optical unit is designed to image radiation with at least two different wavelengths with the same spot size range, in particular with the same spot sizes, in the detector plane.

[0020] The optical device is fundamentally designed to guide radiation from outside the detection device, particularly in the direction of movement in front of the missile, into the detection device, namely into the detector plane, and to image it there. Within the detector plane of the detector device, the detector device has two groups of detection areas, which are sensitive to different wavelengths or wavelength ranges of the radiation.

[0021] This means that radiation of a first wavelength or wavelength range can be detected using detection ranges from a first group of detection ranges, and that radiation of a second wavelength or wavelength range can be detected using detection ranges from a second group of detection ranges. The wavelengths and wavelength ranges are distinct from one another, specifically non-overlapping, so that all wavelengths of the first wavelength range are different from all wavelengths of the second wavelength range. In other words, the different wavelength ranges do not overlap.

[0022] The optical device can be configured to image radiation with at least two different wavelengths with the same spot size range, in particular with the same spot sizes, in the detector plane, wherein at least one spot is imaged blurred in the detector plane, wherein the control device is configured to transform, by means of a transformation algorithm, in particular based on artificial intelligence, the at least one spot imaged blurred in the detector plane and having a first image sharpness into at least one second image sharpness having a greater image sharpness than the first image sharpness.

[0023] The detection device described herein proposes the use of a combination of optical and control elements, i.e., a combination of optical design and image processing algorithms. Specifically, the detection device captures radiation of different wavelengths using the same detector unit, which has different sets of detection areas. In particular, the detection device can be configured for object detection and / or object tracking. The detection device is designed so that the optical unit guides the radiation of different wavelengths or wavelength ranges into the detector plane, with the different spots in the detector plane lying within the same spot size range, i.e., having nearly identical or identical spot sizes.

[0024] However, this results in at least one of the two spots being imaged blurred in the detector plane. To compensate for this blurring or defocus due to the chosen optical design, the control unit is used, on which the described transformation algorithm can be executed. The at least one spot that is imaged blurred in the detector plane is transformed or "re-sharpened" using the transformation algorithm.

[0025] This advantageously makes it possible to image radiation of different wavelength ranges in the detector plane using the same optical system, with comparable spot sizes and thus enabling detection with equally sized fields of view. The resulting blurring is corrected by the control system, i.e., by image processing with the transformation algorithm. In particular, this eliminates the need for particularly complex optical systems, as the error resulting from the blurred imaging of at least one of the spots in the detector plane is at least partially reduced or eliminated by the transformation algorithm, especially one based on artificial intelligence.

[0026] In other words, the described design proposes that the radiation from the different wavelength ranges, which can differ significantly from one another—for example, a combination of first radiation in the LWIR range (long-wave infrared) and second radiation in the MWIR range (mid-wave infrared)—be imaged as uniformly as possible in the detector plane. Since the radiation from the different wavelength ranges passes through the same imaging optics, it is refracted to varying degrees, particularly due to chromatic aberration. Consequently, the diffraction disks described by their respective PSFs (point spread functions) would differ considerably if their areas were minimized simultaneously. The optical system of the detection device described herein, however, is designed in such a way that...designed so that both diffraction disks are approximately the same size, in particular identical.

[0027] The initial image of the detector system can thus be composed of exactly two different wavelength ranges, with the "images" of the individual wavelength ranges (corresponding to the black and white squares of a chessboard) each being convolved with different PSFs. The sub-images are therefore rendered with varying degrees of sharpness, both with respect to the wavelength ranges and within a wavelength range due to optical distortions, particularly towards the edges. This is especially pronounced with wide-angle optics, as edge light falloff must also be expected.

[0028] Due to the described chessboard entanglement, the initial image from the chessboard detector would not be suitable for direct further processing, especially object detection / acquisition and object tracking. Because of the mathematical complexity of the required conversions, it is proposed here to use data-driven methods, which are collectively referred to as "transformation algorithms".

[0029] In one embodiment of the acquisition device, the transformation algorithm can be trained based on provided ideal images and associated blurred images. The training of the transformation algorithm can, for example, be divided into pre-training and fine-tuning. Ideal images for training can be generated synthetically, for example, using a data processing unit. For fine-tuning, real images acquired using a physical acquisition device can be used.

[0030] To generate the blurred images, a generic PSF can be created for both wavelength ranges, for example, a Gaussian distribution modified by appropriate effects such as aberrations, particularly defocus, astigmatism, chroma, vignetting, optical distortion, and the like. This allows the transformation algorithm to be pre-trained, specifically the artificial intelligence to be trained on potential errors and how to compensate for them. In particular, image pairs can be provided: ideal images and blurred images, such as those produced by the optical system due to blurring in the detector plane.

[0031] During fine-tuning, the described pre-trained model or pre-trained artificial intelligence can be specifically fine-tuned to a real-world acquisition device using real-world measurement data, such as collimator images. The fully trained transformation algorithm can then be loaded into signal processing electronics for inference. For example, the ideal coordinator image might correspond to an image in which the light or radiation originates from only one spatial direction or a very limited solid angle. In the ideal image, for instance, only a single area of ​​the acquisition is illuminated.The output image of the optical device of the real detection device can be used as an input image for the transformation algorithm to be fine-trained, whereby the deviation of the output image of the transformation algorithm to the corresponding ideal collimator image, in which only the corresponding detection area is illuminated, defines the training error for the fine-training.

[0032] The training error is thus specified by the combined deviation of the source images from the ideal source images. The individual deviations can be weighted differently; for example, color image deviations can be weighted with a different constant than intensity image deviations. If color images of both channels are used as source images, the weighting can be set to a constant k = 1. The sum of the training errors can, for example, be calculated using the grayscale values ​​of the acquisition areas of the individual images.

[0033] In a further development of the acquisition device, the transformation algorithm can include a neural network. As already described in general terms, the transformation algorithm can be based on artificial intelligence or machine learning, or include such mechanisms. Specifically, a neural network can be used to sharpen the output image from the optical device, which is projected onto the detector device in the detector plane.

[0034] The transformation algorithm can comprise a (deep) neural network with so-called CNN layers. These CNN layers can mimic the function of the human retina and correspond to classical feature images. Such networks can be pre-trained and retrained for the specific task, as described previously. The preprocessing proposed here, similar to the described CNN layers, can be used as a base / input layer for a deep neural network. Alternatively, it can be used for classical image processing, which uses two values ​​per pixel as input, for example, intensity and color temperature. The input layer does not necessarily have to be fully connected, but can be connected only within a limited area, which corresponds, in particular, to the size of the maximum PSF (photospatial field strength).

[0035] The neural network can use the output image on the detector array, specifically in the detector plane of the checkerboard detector, as its input layer. The number of neurons in the network can be equal to the number of pixels, i.e., the detection areas, of the detector array, for example, M rows and N columns. The output layer of the network can output an intensity image and a color image. In contrast to multimedia applications, where color information is often undersampled because the human eye is not as good at spatially locating color information compared to intensity information, it is proposed here to choose the same dimension for both images.

[0036] Based on the previously described pre-training and the optional fine-tuning or fine-training, the transformation algorithm can be trained to reduce at least one aberration, in particular chroma and / or astigmatism. In addition to the optical aberrations, other effects, especially off-axis aberrations, can be considered during fine-tuning, such as distortions and unforeseen effects that may occur, for example, due to manufacturing tolerances of the detector.

[0037] As previously described, the optical device is designed to image the spots of the different radiation with approximately the same size in the detector plane. In one embodiment, the optical device can be designed such that the image onto the detector plane is configured for radiation of a first wavelength, in particular the longer wavelength, and radiation of a second wavelength is adapted to a spot size of the spot of radiation with the first wavelength. In other words, the optical device can be designed primarily for the first wavelength, in particular the larger or longer wavelength, or the wider wavelength range, in particular LWIR, so that the spot of radiation is imaged approximately sharply in the detector plane.The radiation of the second wavelength can then be focused onto the detector plane such that the spot size of the resulting spot corresponds to the spot size of the spot emitted by the radiation of the first wavelength. This means that the radiation of the second wavelength must be defocused accordingly so that the spot sizes are equal.

[0038] As already described, the different wavelengths are distinct wavelengths or wavelength ranges that do not overlap. In one embodiment, the first wavelength can comprise long-wave infrared radiation (LWIR), in particular 8 µm to 15 µm, and / or the second wavelength medium-wave infrared radiation (MWIR), in particular 3 µm to 5 µm. In other words, the optical device can be configured to image long-wave infrared radiation onto the detector plane as the first wavelength and medium-wave infrared radiation onto the detector plane as the second wavelength. The optical device can be designed such that it is suitable for the first wavelength, i.e.,optimized for long-wave infrared radiation, so that radiation with the second wavelength is correspondingly defocused and imaged onto the detector plane, so that the spots of long-wave infrared radiation and the spots of medium-wave infrared radiation are the same with respect to spot diameter or spot size.

[0039] The described control device can be further configured to perform a color transformation of at least two different wavelengths. Advantageously, it is also possible to obtain a single intensity and color image directly, instead of two separate intensity images from the two wavelength ranges. This advantageously allows for the further processing of intensity and temperature characteristics, particularly with pixel-level precision. Pixel-precise intensity conversions from individual color channels to intensity and color signals, specifically using a so-called color transformation known in the visible spectral range, are performed here based on two intensity images in the infrared range, which are not available with pixel-level precision.These are processed here in the sense of a "superpixel" of the checkerboard detector, described below, whose individual gray values ​​are convolved with the respective PSF. Alternatively, the output layer can contain images of both color channels, meaning that no color transformation is performed. The considerations described above regarding the dimensions of the output images apply analogously.

[0040] Furthermore, the detection device may be configured with an optical unit designed to illuminate a group of four detection areas, in particular two detection areas each from the first and second groups, with the spots. This creates a so-called "superpixel" comprising two detection areas from the first group and two detection areas from the second group, particularly arranged crosswise. The optical unit can be designed such that the spot size of each of the two spots illuminates four detection areas. For example, the spot diameter can be twice the pixel length or pixel width. This ensures that no gaps occur between the individual superpixels, as multiple pixels or detection areas are always illuminated by the two spots.

[0041] Basically, two alternatives are possible. In the first alternative, the centers of superpixels can be considered using a dimension (M-1) x (N - 1). These superpixels, as described, are each formed from two crosswise arranged acquisition areas from different groups. While this allows for maximum resolution, the resulting images correlate with each other because the values ​​of the individual acquisition areas are each incorporated into four superpixels or correspondingly arranged evaluation windows. In the second alternative, non-overlapping superpixels can be considered using a dimension (M / 2) x (N / 2). These are correlated only by the overlap of the PSF (Process Field Framework).

[0042] The detection device may further be configured with a control unit designed to evaluate detection areas within a sliding evaluation window, wherein the evaluation window comprises a group of four detection areas, in particular two detection areas each from the first group and the second group. The evaluation window can thus cover or evaluate a previously described "superpixel" in any position. The evaluation window can slide across the detection areas or within the detector plane. This means, in particular, that the evaluation window is not assigned to a fixed grouping of detection areas, but rather that the evaluation window can be positioned over any group of four or a superpixel within the detector plane.

[0043] The sliding evaluation window is particularly useful for object tracking, as there is no signal drop when the spot leaves the evaluation window or the currently illuminated superpixel. Instead, the evaluation window can always be set for the group of four or the superpixel currently illuminated by the two spots. This means that the individual superpixels are not independent of each other; rather, each pixel or detection area can be part of four adjacent superpixels.

[0044] In addition to the described detection device, the invention relates to a missile, in particular a guided missile, which includes a previously described detection device. The missile can, in particular, be configured to detect and track a target signature by means of the described detection device.

[0045] Furthermore, the invention relates to a method for operating a detection device, in particular a previously described detection device, wherein radiation is imaged onto a detector plane of a detector device of the detection device in at least one spot by means of an optical device, wherein the detection device has a phase element which is configured to generate a blur of the at least one spot in the detector plane, wherein, by means of a transformation algorithm, in particular based on artificial intelligence, the at least one spot imaged blurred in the detector plane, having a first image sharpness, is transformed into at least one second image sharpness having a greater image sharpness than the first image sharpness.

[0046] The described method can be implemented in all its details and with all its features on the previously described detection device, or such a detection device can be used to implement the method described herein. All advantages, details, and features described with respect to the detection direction are therefore fully transferable to the method and the missile.

[0047] The invention is explained below with reference to exemplary embodiments and the figures. The figures are schematic representations and show: Fig. 1. A schematic representation of a detection device for a missile according to an exemplary embodiment; Fig. 2 a schematic representation of a detector plane of the detection device of Fig. 1; and Fig. 3. A schematic representation of a detector plane of the detection device of Fig. 1, Fig. 2.

[0048] Fig. Figure 1 shows a schematic diagram of a detection device 1 for a missile (not shown in detail), in particular a guided missile. The missile can detect and / or track target signatures by means of the detection device 1. The detection device 1 includes an optical unit 2 configured to focus radiation onto a detector plane 3 of a detector unit 4 of the detection device 1.

[0049] The optical device 2 is fundamentally designed to image radiation 5, 6 onto the detector plane 3, for example, radiation 5, 6 from a specific wavelength range or with a defined wavelength. Specifically, as illustrated by example, the optical device 2 can image two different types of radiation 5, 6 onto the detector plane 3 of the detector device 4: first, radiation 5 of a first wavelength or from a first wavelength range, in particular long-wave infrared (LWIR), and second, radiation 6 of a second wavelength or from a soft wavelength range, in particular mid-wave infrared (MWIR). The different types of radiation 5, 6 are shown as solid lines for the first type of radiation 5 and as dashed lines for the second type of radiation 6.

[0050] It follows that the radiation 5, 6 is imaged differently by the optical device 2 due to their different wavelengths. The optical device 2 can be specifically optimized for the longer wavelength, namely the first radiation 5, so that it forms a spot 7 (cf. Fig. 2, Fig. 3) can be imaged as optimally as possible in the detector plane 3. The radiation 6 of the second wavelength is adjusted, or the optical device 2 is designed, such that a spot 8 of the second radiation 6 corresponds to the same spot size range as the spot 7 of the first radiation 5. This means that, as for example in Fig. 2, Fig. As shown in Figure 3, the two spots 7 and 8 have almost the same spot size or are exactly the same size. Spot size can be understood, for example, as the spot diameter. For clarity, the two spots 7 and 8 are shown slightly differently, although they could also be exactly the same size.

[0051] The detection device 1 also includes a phase element 9 configured to generate a blur of the at least one spot 7, 8 in the detector plane 3. The phase element 9, for example a cubic phase plate, can be provided as a separate phase element 9 and / or a phase element 9 can be part of an optical element of the optical device 2, for example a mirror and / or a lens. The phase element 9 thus selectively generates a blur of the two spots 7, 8 or of the at least one spot 7, 8 in the detector plane 3, which can then be at least partially compensated for by a transformation algorithm.This allows a stability region to be generated along the optical axis by the phase element 9, since the spots 7, 8 are stable along the optical axis over the stability region and therefore change comparatively little, especially compared to a sharp imaging of the radiation 5, 6 into the detector plane 3.

[0052] As described, a transformation algorithm is executed to increase the image sharpness of at least one spot 7, 8. The detection device 1 includes a control unit 10 designed to execute the transformation algorithm, specifically based on artificial intelligence. The transformation algorithm is intended to transform the spot 7, 8, which is imaged blurred in the detector plane 3 and has a first image sharpness, into a spot with a second image sharpness that is greater than the first.

[0053] In other words, the transformation algorithm is used to sharpen the image of the first radiation 5 and the second radiation 6 in the detector plane 3. Alternatively or additionally, it is also possible to sharpen only a single spot 7, 8, for example, spot 7 of the first radiation 5 or spot 8 of the second radiation 6, by applying the transformation algorithm. As described, the detection device 1 can also be monochromatic, meaning that only radiation 5, 6 from a specific wavelength range or wavelength is imaged onto or processed by the detector device 4.

[0054] The transformation algorithm can operate based on artificial intelligence, for example, machine learning. The transformation algorithm can specifically comprise a neural network. Ideal images and corresponding blurred images can be provided to train the transformation algorithm. In other words, the transformation algorithm is provided with information on which errors caused by the blurred image must be compensated for by the optical device 2 and the phase element 9. The transformation algorithm can thus be trained to compensate for or eliminate the defocus or blur, for example, in spots 7 and 8. For example, synthetically generated images can be provided to the transformation algorithm during pre-training, allowing the transformation algorithm to be pre-trained on any number of synthetically generated images.The synthetically generated images can be produced in particular based on phase information that describes the blur of at least one spot 7, 8 generated by the phase element 9.

[0055] Subsequently, fine-tuning or adjustment can be performed, in which the images are generated by the acquisition device 1 and can be compared, for example, with ideal images. In particular, it is possible to train the transformation algorithm to recognize the aberrations and effects that occur in the real-world acquisition device 1.

[0056] As exemplified in Fig. 2, Fig. As shown in Figure 3, the detector assembly 4 has two groups of detection areas 11 and 12 in detector plane 3, which are arranged alternately in two dimensions in detector plane 3. The detector assembly 4 can therefore also be understood as a bispectral checkerboard detector. For example, detection areas 11 (shown hatched) are sensitive to the first radiation 5, and detection areas 12 (shown in white) are sensitive to the second radiation 6. Alternatively, the detector assembly 4 can also have only identical detection areas 11 and 12.

[0057] In the illustrated embodiment, the optical device 2 is designed and configured such that the spots 7, 8 of the different radiation 5, 6 are imaged in the detector plane 3 as approximately or exactly the same size, and a higher-order pixel or superpixel is illuminated, which comprises two detection areas 11, 12 of the different groups. In other words, the two spots 7, 8 in the detector plane 3 always illuminate two detection areas 11 of the first group and two detection areas 12 of the second group, which together form a superpixel.

[0058] To evaluate the detector array 4, an evaluation window 13 is considered, which corresponds to such a superpixel in terms of size in the detector plane 3. The evaluation window 13 can be evaluated continuously or formed continuously using spots 7, 8 in the detector plane 3. Fig. Figure 3 shows an example of the movement of spots 7 and 8 in the detector plane 3, whereby the evaluation window 13 can slide or be moved together with spots 7 and 8. In particular, during object detection or object tracking, the object emitting radiation 5 and 6, or the corresponding spots 7 and 8, can therefore be tracked without gaps, so that no drop in the signal from the detector device 4 occurs.

[0059] Fig. Figure 3 shows, purely as an example, a movement of spots 7 and 8. Fig. 2 from diagonally in the detector plane 3. Since the two spots 7, 8 are of the same size and each illuminates a superpixel as shown, it is impossible for one of the spots 7, 8 to fall exclusively on a detection area 11, 12 that is not sensitive to this radiation 5, 6. If, for example, the spot size were smaller than or on the order of the size of the detection areas 11, 12, a spot 7, 8 could lie exclusively on one of the detection areas 11, 12 and therefore could not be continuously detected, since this area is not sensitive to the corresponding wavelength.

[0060] By using the sliding evaluation window 13 shown, which always includes a group of four detection areas 11, 12, namely exactly two detection areas 11 and exactly two detection areas 12, both spots 7, 8 of the different radiation 5, 6 are detected in each area of ​​the detector plane 3.

[0061] As already described several times, the exemplary embodiment is in Fig. Figures 1-3 are to be understood as merely examples, since the detection device 1 can also operate monochromatically, i.e., the detector device 4 is only sensitive to a specific wavelength or wavelength range. In any case, the transformation algorithm can be executed by means of the control device 10 to compensate for the blurring generated by the phase element 9.

[0062] If the detector device 4 is operated bispectrally, the stability range generated by the phase element 9 offers particular advantages, since the spots 7, 8 are stable over a larger area along the optical axis. External influences on the detection device 1, such as thermal influences during operation, especially in the case of a cooled detector device 4 during flight, particularly during the operation of a missile with detection direction 1, and / or influences due to manufacturing tolerances and / or assembly tolerances, are therefore less relevant for detection direction 1. Furthermore, the detection device 1 is more robust in this respect due to the phase element 9 and the control unit 10, which executes the transformation algorithm. Reference sign 1 Detection device 2 Optical device 3 Detector level 4 Detector device 5, 6 radiation 7, 8 Spot 9 phase element 10 Control unit 11, 12 Recording area 13 evaluation windows

Claims

Detection device (1) for a missile, in particular a guided missile, comprising an optical device (2) configured to image radiation (5, 6) onto a detector plane (3) of a detector device (4) of the detection device (1) into at least one spot (7, 8), wherein the detection device (1) has a phase element (9) configured to generate a blur of the at least one spot (7, 8) in the detector plane (3), characterized in that a control device (10) is configured to transform, by means of a transformation algorithm, in particular based on artificial intelligence, the at least one blurred spot (7, 8) in the detector plane (3) having a first image sharpness into at least one second image sharpness having a greater image sharpness than the first image sharpness. Detection device (1) according to claim 1 , characterized in that the control device (10) is configured by means of the transformation algorithm to at least partially compensate for the blurring generated by the phase element (9). Detection device (1) according to one of the preceding claims, characterized in that the detection device (1) is configured by means of the phase element (9) to generate a stability region along the optical axis of the detector device (4). Detection device (1) according to one of the preceding claims, characterized in that the control device (10) is configured to transform the at least one spot (7, 8) based on phase information of the phase element (9) by means of the transformation algorithm. Detection device (1) according to one of the preceding claims, characterized in that the phase element (9) is designed as a separate phase plate and / or as a component of an optical element of the optical device (2). Detection device (1) according to one of the preceding claims, characterized in that the phase element (9) has an asymmetrical geometry, in particular a freeform surface. Detection device (1) according to one of the preceding claims, characterized in that the detector device (4) has two groups of detection areas (11, 12) which are designed to be sensitive to different wavelengths and / or wavelength ranges of the radiation (5, 6), wherein the optical device (2) is designed to image radiation (5, 6) with at least two different wavelengths with the same spot size range, in particular with the same spot sizes, in the detector plane (3). Detection device (1) according to one of the preceding claims, characterized in that the transformation algorithm is trained on the basis of provided ideal images and associated blurred images. Missile, in particular guided missile, comprising a detection device (1) according to one of the preceding claims. Method for operating a detection device (1), in particular a detection device (1) according to one of claims 1 to 8, wherein radiation (5, 6) is imaged onto a detector plane (3) of a detector device (4) of the detection device (1) in at least one spot (7, 8) by means of an optical device (2), wherein the detection device (1) has a phase element (9) configured to generate a blur of the at least one spot (7, 8) in the detector plane (3), characterized in that by means of a transformation algorithm, in particular based on artificial intelligence, the at least one spot (7, 8) imaged blurred in the detector plane (3) and having a first image sharpness is transformed into at least one second spot (7, 8) having a greater image sharpness than the first image sharpness.

Images