Method and system for monitoring mobile space objects

Abstract

The invention relates to a method for monitoring mobile space objects, the method being implemented by a monitoring system (100), the method comprising the following steps: - slidingly acquiring at least one stack (12, 12') of image frames of space according to a known trajectory on the celestial vault, each stack of frames comprising at least three frames, each frame being recorded with an exposure time of less than 1 second; - detecting, in the field of view, a plurality of stars (14, 55), preferably several tens, in the at least one stack; - estimating the coordinates of the detected stars; - identifying the detected stars with a list of reference stars extracted from a stellar catalog, the stellar catalog comprising several thousand reference stars; - establishing a coordinate relationship linking the celestial coordinates to the coordinates in the frames of the stars present in the list of reference stars; - detecting at least one mobile space object (15, 56) in at least one stack and estimating the dated frame coordinates of the at least one mobile space object; - converting the dated frame coordinates of each detected mobile space object into coordinates on the celestial vault by the coordinate relationship.

Description

[0001] Method and system for monitoring mobile space objects

[0002] technical field

[0003] The present invention relates to a method for monitoring moving space objects. The invention also relates to a system for monitoring moving space objects implementing such a method.

[0004] The field of the invention is, without limitation, that of space surveillance, and in particular that of the fine optical estimation of the dated position of objects in Earth orbit.

[0005] This document considers different types of objects present in space.

[0006] The term "space object" refers to any type of object present in space, including stars, asteroids such as near-Earth objects, and artificial space objects in Earth orbit (R.SO, resident space objects according to English terminology).

[0007] The term "moving space object" refers to space objects, excluding stars.

[0008] State of the art

[0009] Space surveillance has become a critical issue given the increasing number of satellites, the benefits derived from their operation (Earth observation, positioning, telecommunications, etc.), the proliferation of space debris, and other factors. Optical and radar measurements are continuously performed around the Earth to detect space-borne objects (SBOs), such as active or deactivated satellites, launch vehicle stages, and debris, visible from each location, and to estimate their state—that is, their positions and velocities at a given time, within a given frame of reference. In this document, we consider the state to be their projection parallel to the line of sight, a quantity accessible through passive optical measurements. The quality of these measurements is a key parameter for their use.They allow the detection of new RSOs as well as the estimation of orbit from their states, for example to be able to manage the risks of collision or detect maneuvers.

[0010] Space surveillance also includes the search for near-Earth objects, such as asteroids that pass close to Earth, also called NEOs (near-Earth objects) according to English terminology.

[0011] Space surveillance systems aim, in particular, to maintain the most comprehensive list possible of RSOs (Remotely Seen Objects) and their movements, in order to identify them on each pass and monitor the evolution of their parameters. Excluding maneuvers, which are generally of small amplitude, their deterministic trajectory, governed by the laws of space mechanics, is entirely determined by a set of six parameters. Ground stations therefore aim to measure a few dated positions of RSOs in order to estimate these parameters. It should be noted that RSOs are so distant that most are "unresolved" by ground-based observation systems; that is, their image (like that of a star) is indistinguishable from that of a point source, except for its movement relative to the background of stars.

[0012] For certain specific cooperative satellites, that is to say, equipped with laser retroreflectors, radio frequency transmitters / receivers, referred to here as "reference satellites", ground-based means have been deployed to measure their position with centimeter-level accuracy.

[0013] For other satellites, measurement accuracy was not a priority for state-of-the-art space surveillance systems, for several reasons:

[0014] - Exact orbital calculations are complex and computationally expensive. Simplified algorithms of limited precision (on the order of kilometers), taking into account the effect of higher-order perturbations, are still used and are sufficient for many needs.

[0015] - Due to the low density of the space population, imprecise orbits were sufficient to identify RSOs, predict their future position and manage the risks of collisions.

[0016] - It is very difficult to cover the entire celestial sphere with high spatial and temporal resolution. The need to cover a wide field of view implies the difficulty of covering it precisely at a fixed data rate. The measurement rate, that is, the number of square degrees of field per second and the sampling resolution, was limited by the performance of the image sensors and computers.

[0017] However, with the very strong growth in the number of R.SOs in space, the need for precision is becoming more demanding.

[0018] Description of the invention

[0019] One aim of the present invention is to propose a method and system for monitoring mobile space objects, enabling such objects to be monitored more precisely.

[0020] In particular, one aim of the present invention is to provide a method and a system for monitoring moving space objects, enabling the detection of a potentially tenuous moving space object against a strong background.

[0021] Another objective of the present invention is to provide a method and system for monitoring moving space objects, enabling the precise localization of a moving space object on the celestial sphere at a given time.

[0022] At least one of these goals is achieved with a method for monitoring moving space objects, a moving space object including an artificial orbital object or a near-Earth object, the method being implemented by a ground-based or space-based monitoring system comprising at least one optical tube or optical lens, at least one camera, a computing unit and a control unit, the method comprising the following steps:

[0023] - pointing the optical tube along a known trajectory on the celestial sphere,

[0024] - sliding acquisition, by the camera, of at least one stack of image frames of space according to a field of view, each stack of frames comprising at least three frames, each frame being recorded with an exposure time of less than 1 second,

[0025] - detection, within the field of view, of a plurality of stars, preferably several dozen, in at least one stack of image frames, - estimation of the coordinates, in at least one frame and at a given time, of the stars detected in the field of view,

[0026] - identification of detected stars using a list of reference stars previously extracted from a stellar catalogue, the list comprising several thousand reference stars,

[0027] - establishment of a relationship, called a coordinate relationship, linking the celestial coordinates to the coordinates in the frames of the detected stars present in the list of reference stars,

[0028] - detection of at least one moving spatial object in at least one frame stack and estimation of its dated frame coordinates,

[0029] - conversion of the dated frame coordinates of each detected moving spatial object into coordinates on the celestial sphere using the coordinate relation, and

[0030] - transmission of coordinates on the celestial vault to the control unit or another computing unit.

[0031] The term "sliding acquisition" refers to the acquisition of frames, or images, according to the pointing path. In other words, during the sliding acquisition of the frame stack, a pointing path is applied to the optical tube.

[0032] Advantageously, the pointing trajectory is substantially linear and at uniform speed in the camera's image plane.

[0033] According to one embodiment, the pointing trajectory can correspond to a stellar pursuit.

[0034] According to another embodiment, the pointing trajectory can correspond to a pointing trajectory with average speed weighted between the apparent speeds of the stars and the expected moving objects.

[0035] Alternatively, the pointing trajectory can be zero, meaning the optical tube is in fixed aiming. Advantageously, the method can include the parameterized modeling of an image spot of a spatial object in the field of view, the image spot having a defined shape and motion, the spatial object comprising a star or a moving spatial object.

[0036] The parameterized image spot model can be used during the star and / or moving space object detection steps, and / or during the fine determination of the coordinates of space objects.

[0037] According to one embodiment, the detection of at least one space object may include the following steps, the space object comprising a star or a moving space object:

[0038] - choice of at least one exploration path for the frame stack, the exploration path being defined as a shift dx(k), dy(k) applied to each pixel (x, y) in the frame with index k,

[0039] - calculation in the frame stack, along each exploration path, of a test statistic map for a plurality of pixels (x, y) and offsets dx(k), dy(k) of interest, and

[0040] - possible extraction of spatial objects whose test statistic is greater than a threshold value.

[0041] It is possible to iterate these steps over a group of exploration trajectories of interest and to extract spatial objects only at the end of the exploration.

[0042] For example, for stars, the sidereal trajectory can be followed with possible variations to integrate a tracking defect of the mount or field rotation.

[0043] For known spatial reference systems (SRS), the expected trajectory can be followed with a margin of variation; for unknown SRS, a large list of potential velocity vectors can be used, these vectors covering a region of interest in the two-dimensional space of apparent velocities. Advantageously, the exploration trajectory can correspond to a uniform linear motion, resulting from the combination of the motion of the spatial objects and the pointing trajectory, with the motion of the spatial objects and the pointing trajectory being substantially linear over the duration of the frame stack.

[0044] According to one embodiment, the method further comprises a step of determining the coordinates of the detected space object, the space object comprising a star or a moving space object, the determination of the coordinates comprising at least one of the following steps:

[0045] - approximation of the spatial object's position by its integer coordinates (x, y) corresponding to the local maxima of the previously calculated test statistic map,

[0046] - estimation of the sub-pixel coordinates of the detected spatial object through finer positioning of the detected maximum,

[0047] - refinement of the position of the detected spatial object by optimizing the detection signal-to-noise ratio, with a classic iterative algorithm not based on the quantification of the parameters sought.

[0048] According to one embodiment, the method according to the invention further comprises a step of finely determining the coordinates of a detected space object, the space object comprising a star or a moving space object, the finely determining comprising at least one of the following steps:

[0049] - modeling the displacement of the moving space object and the observer during the light propagation time between the moving space object and the camera,

[0050] - modeling of the effect of annual aberration, experienced by stars but not by moving space objects,

[0051] - modeling, for a scanning sensor, of the effects of variation in the field at the actual time of measurement,

[0052] - Modeling of optical tube distortion using a polynomial of at least third order to establish the coordinate relation, - Modeling of chromatic effects by taking into account a correction term based on the spectral distribution of the spatial object,

[0053] - modeling the effects of variation in the field of atmospheric refraction, of diurnal aberration by using a polynomial of at least third order to establish the coordinate relation,

[0054] - modeling of the parallax effect induced by atmospheric refraction.

[0055] These steps allow, alone or in combination, a so-called fine determination of the coordinates of at least one spatial object.

[0056] The coordinates of a moving spatial object include its position.

[0057] Advantageously, the detection of at least one moving space object may also include determining its velocity in the observed plane, its average magnitude and / or its photometric variation.

[0058] The average magnitude can be determined in particular by means of a magnitude relationship established using reference stars.

[0059] Depending on the embodiment, the list of reference stars can be determined from a star catalogue by:

[0060] - the selection of stars based on at least one of the following characteristics: magnitude, temperature, multiplicity, variability, distance to neighboring stars,

[0061] - updating the coordinates of the stars in the list with the Earth's orbital parameters and the characteristics of the stars (distance, proper motion).

[0062] The number of stars detected and the richness of the reference list are such that several dozen stars are identified in the stack of images.

[0063] The sublist of reference stars, obtained by intersecting the stars in the reference star list extracted from the catalog for most of the celestial sphere and the stars detected in the field of view, preferably includes several dozen stars, or even preferably more than a hundred. This requires a reference star list with a very large number of stars, depending on the sensor's field of view.

[0064] The reference list can be updated to a date close enough to the observation to take into account the position of the stars at that date.

[0065] Advantageously, the method according to the invention can include a preprocessing step of each stack of frames, prior to the step of detecting a plurality of stars in the field of view.

[0066] Depending on the embodiment, the preprocessing step may include one or more of the following steps:

[0067] - Filtering of frames in the frame stack to eliminate a static background,

[0068] - pixel grouping of the frames,

[0069] - reduction of the field of view, and / or

[0070] - correction of frames on the calibration database of the monitoring system.

[0071] According to one embodiment, the method includes fixing the camera integration time for each frame so that during the acquisition time, the movement of a spatial object covers less than 10 pixels of a frame in the frame stack, and preferably less than one pixel.

[0072] Advantageously, the method according to the invention can include a precise timestamping of a frame from the stack of frames, preferably at a central instant of the stack.

[0073] This precise timing is on the order of milliseconds, with a typical variation, depending on the required accuracy, between ten milliseconds and one-tenth of a millisecond. The method according to the invention may further include a preliminary step of determining the terrestrial coordinates of the monitoring system.

[0074] The method according to the invention may also include a preliminary or regular calibration step of the monitoring system, and therefore in particular of the telescope and the computing unit or units, by observation of at least one reference satellite.

[0075] According to another aspect of the same invention, a system for monitoring mobile space objects is proposed, comprising:

[0076] - at least one optical tube or lens, configured to be pointed along a known trajectory on the celestial sphere,

[0077] - at least one camera, configured for the sliding acquisition of a stack of frames of space according to a field of view produced by the optical tube, the frame stack comprising at least three frames, each frame being recorded with an exposure time of less than 1 second,

[0078] - a unit of calculation,

[0079] - a control unit, the system being configured to implement a method for monitoring mobile space objects according to the invention.

[0080] Such a system, also called a ground or space station, is a combination of an opto-mechanical device (comprising one or more optical tubes or lenses focusing a star field, or field of view, optionally the mount associated with each tube), an image-capture system (comprising one or more cameras), and acquisition, processing, and control means (comprising one or more computers, including RAM, possibly graphics processors, storage and communication devices), implementing the algorithms of the method according to the invention. The optical tube operates, in particular, over a wide range of wavelengths, from ultraviolet to infrared.

[0081] The term "tube" is generally reserved for astronomical instruments. They are sometimes replaced by "lenses," a term used in photography.

[0082] Advantageously, the system according to the invention may further include one or more spectral filters. These can be implemented, for example, using an insertion module.

[0083] One or more spectral filters can, for example, be used to reduce the sky background in degraded conditions (for example, during daytime observations).

[0084] The method and system for monitoring moving space objects can be implemented to perform precise orbitography from detailed measurements of positions and velocities. This allows for better prediction of a satellite's trajectory, for example, paving the way for numerous applications:

[0085] - maintaining a private catalogue, including items not listed in online catalogues,

[0086] - early detection of satellite maneuvers,

[0087] - precise piloting from the ground for on-orbit servicing missions,

[0088] - more precise assessment of collision risks to minimize the cost of avoidance maneuvers.

[0089] Description of the figures and methods of implementation

[0090] Other advantages and features will become apparent upon examination of the detailed description of examples, which are by no means exhaustive, and the accompanying drawings on which:

[0091] - [Fig. 1] Figure 1 is a schematic representation of a non-limiting example embodiment of a mobile space object monitoring system according to the invention,

[0092] - [Fig.2] Figure 2 illustrates the principle of generating a pseudo-long-exposure image obtained in the process according to the invention, - [Fig.3] Figure 3 is a schematic representation of an example of sampling a moving point by a matrix image sensor,

[0093] - [Fig. 4] Figure 4 is a schematic representation of the steps in the monitoring process according to one embodiment of the invention, and

[0094] - [Fig.5] Figure 5 illustrates parasitic effects induced during the passage of light through the atmosphere.

[0095] It is understood that the embodiments described below are by no means exhaustive. In particular, variants of the invention may be conceived comprising only a selection of the features described below, isolated from the other features described, if this selection of features is sufficient to confer a technical advantage or to differentiate the invention from the prior art. This selection includes at least one preferably functional feature without structural details, or with only a portion of the structural details if this portion alone is sufficient to confer a technical advantage or to differentiate the invention from the prior art.

[0096] In particular, all the variants and embodiments described can be combined with each other if there are no technical obstacles to this combination.

[0097] In the figures, elements common to several figures can retain the same reference.

[0098] Figure 1 is a schematic representation of a non-limiting example of a system 100 for monitoring mobile space objects, which can be implemented within the scope of the present invention. The system 100 according to the example shown can, in particular, be implemented in a method for monitoring mobile space objects according to the invention.

[0099] The system 100 according to the invention comprises an opto-mechanical system 10, that is, a telescope, for observing the sky from a space or ground station. In the example shown in Figure 1, this is a ground station. The telescope 10 comprises an optical tube. The system 10 may also comprise one or more mounts and several optical tubes.

[0100] The optical tube includes optics for focusing a star field, or field of view. The optical tube is pointed by the mount along a known trajectory T on the celestial sphere. One or more image sensors, such as cameras, record a series of image stacks of the sky within this field of view.

[0101] The image sensor, or image sensors, is preferably a CMOS (Complementary Metal Oxide Semiconductor) camera. These parallel architecture cameras allow for a combination of wide field of view and high readout speed. Even though CMOS sensors can have higher readout noise than CCD cameras, this noise can still be lower than the background noise.

[0102] The acquisition with the camera or cameras is a so-called sliding acquisition, due to the realization of the trajectory T by the tube, which can be fixed.

[0103] The acquired images include stars and any moving space objects, such as solar solar arrays (SSAs). The 14 stars exhibit a deterministic apparent motion R, driven by Earth's rotation (15 arcseconds per second); other motions of lesser magnitude are also present (including annual and diurnal aberration, annual parallax, and proper motion). The 15 moving space objects have an apparent velocity S that varies in direction and magnitude, from approximately 0 (objects near geostationary orbit) to 1 degree per second (objects in low Earth orbit).

[0104] Each image stack, or image frame stack, can be transmitted to a computer, or computing unit 13. This computer detects the stars and any moving space objects present and extracts their image coordinates (x, y) into a reference frame whose precise date is known. These two categories of objects can be distinguished by their velocities, as explained above. From a list of reference stars extracted from a star catalog G, a correspondence is established between the image coordinates (x, y) of the stars and the celestial coordinates (a, 5). The image coordinates (x, y) of the space objects can then be converted into celestial coordinates (a, 5).Similarly, in addition, the correspondence between the photometric levels of the pixels and the stellar magnitude scale can be established using the stars in the reference list, which constitutes a magnitude relationship, and thus the magnitudes of moving space objects can be estimated. It is also possible to estimate the apparent velocity of the objects. This information is transmitted to a downstream computer, or computing unit. This downstream computer can, for example, perform an orbit estimation.

[0105] System 100 is governed by a controller 19, or control unit, configured to coordinate the operations performed by the mount, the camera(s) and the computing units 13, 11.

[0106] The computing unit 13 can be located near the telescope 10. It can be connected, for example, by a fiber optic link, or any other high-speed link.

[0107] The downstream computing unit 11 can be located away from the telescope 10. A high-speed link is not required.

[0108] For example, for a space system, the computing unit 13 must be on board, and the downstream computing unit 11 can be on the ground.

[0109] The computing units 13 and 11 can also be combined into a single unit. The control unit 19 can also be physically combined with the computing unit(s) 13 and 11.

[0110] These units may include one or more computers, including RAM, possibly graphics processors, storage and communication devices.

[0111] According to one embodiment, a massively parallel computation 13a is performed on a large number of pixels and potentially velocities in order to detect spatial objects of previously unknown position and velocity. In some cases, the velocity may be known (for example, the case of stars).

[0112] This calculation can be performed, for example, by at least one graphics processing unit (GPU) to determine indices (position, velocity) on a quantized scanning grid. In another embodiment, the detections made during processing 13a are processed in a second pass 13b by a more refined calculation from the initial frame stack, restricting the analysis to local areas around the previously made detections. Finer effects, such as chromatic aberration from refraction or the tube, or the parallax of moving space objects 15 relative to stars 14, can be included, reducing the measurement error to a few tens of milliarcseconds.Thus, from a coordinate relation identified on the stars present in the field and an interpolation polynomial to take into account the variations in the field (variation of refraction with the zenith angle, distortion of the optical tube), a precision of less than one arcsecond can be obtained.

[0113] The position (a, 5) of the stars depends slightly on the position of the Earth 27 on its solar orbit, on effects due to its rotation speed (around the Sun 28 and on itself) and on the orientation of its north-south axis in space.

[0114] The image processing performed by the calculator 13 according to the different embodiments will be described in more detail below.

[0115] Advantageously, the system according to the invention may also include one or more spectral filters. These can be implemented, for example, using an insertion module.

[0116] These spectral filters can, for example, be used to reduce the sky background in degraded conditions (for example, during daytime observations).

[0117] Spectral filters, for example in the red - near-infrared and / or in the blue - green, can also be used to estimate a color index on a detected moving space object and on stars, in order to better constrain the chromatic model.

[0118] The monitoring system according to the invention, for example according to the embodiment shown in Figure 1, can be implemented in a method for monitoring moving space objects according to the present invention. The method according to the invention comprises a step of pointing the optical tube by the mount along a known trajectory on the celestial sphere.

[0119] Preferably, the pointing trajectory is substantially linear and at uniform speed in the camera's image plane.

[0120] The pointing trajectory can correspond to a stellar pursuit where the telescope mount compensates for the Earth's rotation.

[0121] The pointing trajectory can also correspond to the pursuit of a moving spatial object, such as an R.SO.

[0122] The pointing trajectory may still correspond to a pursuit on a weighted average speed between the stars and a moving space object.

[0123] Alternatively, the pointing trajectory can be zero, that is, the optical tube is in fixed aiming.

[0124] During an acquisition step, at least one image stack of the space according to a field of view is acquired. Each image frame stack comprises at least three image frames.

[0125] The image stack can also be called a frame stack, or image frame stack.

[0126] Each image stack or frame comprises a series of consecutive images where the viewing direction, or pointing direction, varies almost continuously between each image over time. There are no temporal or spatial jumps between images. An image stack constitutes a three-dimensional array of values, I(x, y, k), with, for each image, the index k and the pixels (x, y).

[0127] A plurality of frame stacks can be acquired for a plurality of pointing trajectories. These trajectories can be linked, for example, by performing successive aiming directions to create a monitoring loop for a certain time.

[0128] Each image frame is recorded with an exposure time of less than 1 second. These are therefore called "short-exposure" frames, due to their short exposure time. Images from a stack of short-exposure images can be used to generate a pseudo-long-exposure image afterward.

[0129] This pseudo long-exposure image can be used to illustrate the combined effect of the proper motion of space objects, the telescope pointing trajectory during acquisition, and the multiple stack exploration trajectories selected during processing, as explained below.

[0130] Figure 2 illustrates the principle of generating a pseudo-long-exposure image from a stack of 12 (temporal, according to k) images. Each image comprises an (x, y) grid of pixels. The images are acquired in "short exposure" mode (i.e., with an exposure time of less than 1 second) while the telescope follows a predetermined pointing trajectory. The observed field includes 14 stars and potentially at least one moving space object 15, such as a solar solar array (SSA).

[0131] Calculating the so-called pseudo long-exposure image involves summing the image frames of the stack 12 by applying a certain offset dx(k), dy(k) 13 to each frame.

[0132] The image stack 12 can be acquired using different sliding acquisition modes, depending on the chosen pointing trajectory. Regardless of the pointing trajectory chosen, it is possible to calculate numerous pseudo-long-exposure images afterward according to the chosen offset, as illustrated at the bottom of Figure 2.

[0133] Indeed, the 13-frame offset can be achieved either during acquisition, by selecting the pointing trajectory, or afterward during the calculation of the pseudo-long-exposure image. In the first case, the offset is fixed and unique during acquisition. In the second case, the offset can be adjusted at will during processing.

[0134] Figure 2(a) shows the pseudo-long-exposure image obtained with zero offset, symbolized by arrows 16, applied to stack 12 while the telescope is in sidereal tracking. In this case, the telescope corrects for the Earth's rotation. The fixed stars form 14' spots on the pseudo-long-exposure image. The moving SAR appears as a 15' trail. Figure 2(b) corresponds to the pseudo-long-exposure image calculated with zero offset, symbolized by arrow 17', when the telescope is tracking the SAR, resulting in another 12' stack of frames. On the sensor, the SAR produces a 15' spot, and the stars form elongated 14' trails.

[0135] It should be noted that a similar pseudo long-exposure image can be produced with stack 12 (i.e., with the telescope in sidereal tracking) by shifting the frames by 17 during summation.

[0136] Figure 2(c) corresponds to the pseudo long-exposure image calculated in an intermediate case.

[0137] According to a first possibility, this intermediate case can be obtained with the telescope in stellar tracking (the frames are then aligned as in the left stack 12) and a shift according to arrow 18.

[0138] According to a second possibility, the telescope is tracking the R.SO, producing a stack of 12' images, and an 18' offset is applied to this 12' stack.

[0139] According to yet another possibility (not shown), the pointing trajectory follows the average speed between the sidereal speed and the speed of the R.SO, and the offset during summation is zero.

[0140] In the case of Figure 2(c), the two types of objects produce 14', 15' tracks of identical length. The directions of movement of the R.SO spot on the one hand and the star spots on the other are reversed.

[0141] In the example illustrated in Figure 2(c), the length of the traces corresponds to half that of the traces in Figures 2(a) and 2(b), respectively.

[0142] The acquisition of a stack of frames 12, combined with the calculation of a pseudo-long-exposure image with a shift 13, allows, for example, while the telescope is in stellar tracking mode, the synthesis of both a pseudo-long-exposure image of the stars (simple sum of the frames along 16, image in Figure 2(a)) and a pseudo-long-exposure image tracking the SAR (oblique sum of the frames with a suitable shift 17, image in Figure 2(b)). This allows for better detection of both the SAR and the stars. Furthermore, in the image in Figure 2(a), the direction of movement of the SAR cannot be determined from the pseudo-long-exposure image alone, or from the long-exposure image that would have been obtained with a single frame associated with the total integration time. It is nevertheless possible to determine it from the short-exposure images by producing the pseudo long-exposure image with a shift corresponding to the speed of the SAR to form the image of Figure 2(b).This allows us to identify the speed and therefore the type of objects in the image, including in a case like that leading to the image in Figure 2(c) which can be obtained by a combination of pointing and offset trajectories applied.

[0143] As explained above, the pseudo long-exposure image in Figure 2(c) can be obtained in at least three ways: with a pointing trajectory following the stars or the SOR, and the associated offsets of 18 or 18' when processing the stack, or with a pointing speed equal to the average of the two previous pointing speeds, and a zero offset.

[0144] Advantageously, in the latter case, the apparent velocity of image spots, formed by moving sources such as SORs or stars, is halved on the camera. Indeed, the overall performance of the monitoring system depends on the ability to locate both SORs and reference stars. Tracking one of these objects can lead to an inaccurate estimation of the other, primarily due to the elongation of the spot introduced by the movement of an object within each frame. With average velocity tracking, possibly weighted by the characteristics (number, magnitude) of the objects (stars, SORs) expected in the field, the apparent velocities of the star and the SOR are jointly reduced, and intra-frame displacement is minimized. Reducing the size of the image spot increases measurement accuracy. It also increases the integration time, leading to a gain in sensitivity.As a result, the number of frames to be processed can be reduced, and the signal-to-noise ratio on these frames can be increased.

[0145] During the acquisition stage, the camera integration time for each frame is fixed so that during the acquisition time, the movement of a spatial object covers less than ten pixels of a frame in the frame stack, and preferably less than one pixel.

[0146] The integration time can be determined based on prior knowledge of the moving space objects to be detected or observed. For example, for slow-moving targets, integration times that are not too short, representing a compromise between a good signal-to-noise ratio and perturbations induced by turbulence, can be chosen. The faster the moving space objects to be observed or detected, for example, those in lower orbits, the more beneficial it is to reduce the integration time, within the limits of the camera and computing power.

[0147] Figure 3 is a schematic representation of an example of sampling a moving point by a matrix sensor, such as a camera.

[0148] As illustrated in Figure 3(a), for a single image or frame, the Cartesian geometry of the pixels allows, on the one hand, the measurement of an average value on each pixel (represented by the hatched area 31) and, on the other hand, the performance of sampling represented by the disks 32 designating the locations where these average values ​​are measured. The sampling interval then corresponds to a square such as that represented by the dotted square 33.

[0149] In Figure 3(b), we consider the case of a star 34 moving at an angle of 45° relative to the sensor's pixel grid, the displacement being indicated by arrow 35. The 45° angle is an example chosen to simplify the demonstration. Of course, the angle relative to the grid can have other values. If the star's velocity is known, it is possible, by adjusting the sensor's integration time, to perform sampling when the star's displacement distance is equal to half a pixel diagonal. It is then equivalent to moving the sensor in front of the stationary star, which amounts to sampling it on an offset pixel grid, this offset grid being represented by rings 32'. Combining these two grids over two successive frames ultimately yields an oblique square grid with a pitch half the size of the sampling pitch in the case of Figure 3(a).This reduced sampling step is represented by the dashed square 33' in Figure 3(b). Moving the observed spatial object in front of the sensor then increases the sensor's sampling resolution. This technique can be implemented in several ways in the method according to the invention using image stacks, and in particular with a sliding acquisition:

[0150] - exploitation of the natural displacement of a stellar orbital refractor (SOR), even if unknown, in standby mode with stellar tracking,

[0151] - application of a known displacement to improve star sampling, for example such as in Figure 3 of half a pixel at 45°,

[0152] - in the case where the speed of the RSO is too high to constrain the elongation of the image spot to a low value in one direction, it is possible, by choosing the pointing speed during two successive sliding acquisitions, to force the elongation of the spot in two orthogonal directions judiciously oriented with respect to the pixels in order to measure alternately and precisely each of the two components.

[0153] At least one frame in the frame stack is accurately time-stamped, preferably at a central point in the stack. This time-stamping can be achieved by TTL synchronizing the camera to a reference signal, for example, from a GPS sensor. Thus, the recorded images are correctly time-stamped using a Coordinated Universal Time (UTC) reference.

[0154] A timing on the order of milliseconds makes it possible to obtain, on an object of the type of positioning satellite [global navigation satellite system, GNSS), moving at about 40 arcsec / s and for this single error station, an accuracy of a few tens of milliarcseconds.

[0155] In Figure 1, dating is illustrated by a clock symbol.

[0156] According to some embodiments, the process according to the invention includes a preprocessing step for each stack of frames.

[0157] The preprocessing stage may include, on the one hand, steps or actions to correct the image stacks. The frames in the frame stack may be filtered to eliminate a static background. Reducing the background allows, in particular, an increase in the detection limit magnitude.

[0158] A pixel binning technique can be applied to the frames. Pixel binning can be achieved, for example, by keeping the maximum pixels, or pixels correlated by the kernel of a suitable filter, or by keeping the average pixel value.

[0159] The images in the image stack can be corrected based on the monitoring system's calibration database, allowing for the correction of disparities between pixels.

[0160] On the other hand, preprocessing allows each stack of images to be reduced.

[0161] It is possible, in particular, to reduce the field of view. Reducing the field to a region of interest or binning allows processing on the fastest SARs or the widest fields.

[0162] Preprocessing the image stack allows us to eliminate certain unnecessary signals (by removing the background or transient events). It also speeds up processing (by reducing the field of view or grouping pixels) or corrects calibrated defects (pixel gain mapping).

[0163] During a processing phase of the method according to the invention, the image frame stack(s) are processed to detect any moving spatial objects in these images. In some embodiments, the processing phase comprises the following steps.

[0164] During a modeling step, an image spot of a spatial object in the field of view is represented by a parameterized model. The image spot has a defined shape and motion. The spatial object could be a star or a moving object.

[0165] The image spot can be modeled by a spot of known shape, for example of Gaussian shape, of width / , moving in a uniform rectilinear motion of speed v passing through a position p. The parameters p, v and / are then sought.

[0166] When a small optical telescope is used and for relatively short periods, the image spot of a space object is a spot in uniform rectilinear motion for the pointing trajectories considered. The space object is characterized by a position (xo, yo) at an arbitrary time, a velocity (u, v), and a light intensity (expressed as an energy level deposited on the pixel), which, after camera calibration, can be expressed as a number of photons N per unit time.

[0167] The coordinates, in the image plane of each frame, of stars present in the field of view are estimated.

[0168] To achieve this, a plurality of stars, preferably several dozen, are detected in the field of view. This detection can be carried out on a suitable pseudo-long-exposure image (fixed stars) or by searching along a suitable exploration trajectory.

[0169] The search, following an exploration trajectory, is carried out as described below with reference to Figure 4.

[0170] Stars can be distinguished from other space objects by their specific apparent motion.

[0171] The coordinates of the stars are represented in the image plane by the indices of the corresponding pixels (two pixel indices, which may possibly be interpolated).

[0172] The detected stars are then identified using a list of reference stars, extracted from a stellar catalogue. This list includes a very large plurality of stars, preferably several thousand, so that several dozen stars from this plurality can be used in each acquisition field.

[0173] To perform this step, the stars detected in the field of view are identified using a reference star list and appropriate software familiar to those skilled in the art. Their positions in the image are then correlated with their coordinates from the star list. Celestial coordinates may include, for example, right ascension and declination in the International Celestial Reference System (ICRS).

[0174] The list of the plurality of reference stars can be established by selecting stars based on at least one of the following characteristics: magnitude, temperature, multiplicity, variability, distance to neighboring stars.

[0175] In the process according to the invention, a relationship, called a coordinate relationship, linking celestial coordinates of the reference stars to the coordinates of the stars in the image frames is then established.

[0176] In one embodiment, establishing the coordinate relationship between pixel and celestial coordinates can be achieved by taking into account one or more spectral characteristics of the observed stars, such as temperature, a color index, or a magnitude difference between two spectral bands. These spectral characteristics are also provided by a star catalog and allow the establishment of the coordinate relationship to be refined by a chromatic refraction model.

[0177] Advantageously, using a large number of reference stars in establishing the coordinate relation allows for the precise fitting of an interpolation polynomial of sufficient order. It also allows for the averaging, within the field, of atmospheric turbulence residuals, representing one of the most difficult stochastic perturbations to correct.

[0178] The list can be updated by updating the coordinates of the stars in the list with Earth's orbital parameters and characteristics from the catalog of these stars, at a time sufficiently close to the observation. To do this, the list of stellar reference stars is adjusted prior to the observation, at a date close to the time of observation or detection, using all known deterministic elements, such as those resulting from the proper motion of the stars or the Earth (stellar parallaxes, aberrations of light). During a detection step, one or more moving space objects are detected in at least one frame stack.

[0179] According to one embodiment, to perform this detection, at least one exploration path of the frame stack is chosen. The exploration path is defined as an offset dx(k), dy(k) applied to each pixel (x, y) in the frame with index k.

[0180] Preferably, the exploration trajectory corresponds to a uniform linear motion, or its approximation over an entire grid, resulting from the combination of the motion of spatial objects and the pointing trajectory, the motion of spatial objects and the pointing trajectory being substantially linear over the duration of the frame stack.

[0181] The pixels of interest, forming the exploration path, may exclude, in particular, pixels close to the edge of the frames because image processing is not applicable there due to the offsets induced by the exploration path.

[0182] In the frame stack, a test statistic map is calculated for a plurality of pixels (x, y) and dx(k), dy(k) offsets of interest along each scan path.

[0183] From the statistical test map, spatial objects whose test statistic is greater than a threshold value can be extracted.

[0184] The test statistic map can be obtained by dividing the sum, on each frame k of the frame stack, of the intensity of each pixel I(x, y, k) by the sum of the inverses of the standard deviations of the fluctuations associated with the pixel at I(x, y, k). This is equivalent to calculating a signal-to-noise ratio on a pseudo-long-exposure image.

[0185] Advantageously, the test statistic map can also be obtained by appropriate filtering, in particular by summing along the exploration path of each frame k the intensity of pixels I(x + dx(k), y + dy(k), k) on the shape and position of the expected image spot, with inverse signal-to-noise weighting and ad-hoc normalization.

[0186] This calculation will be described in more detail with reference to Figure 4. It is possible to iterate these steps on a group of exploration trajectories of interest. To do this, it would ultimately be necessary to identify the maximum values ​​of the test statistic for all the explored trajectories. To limit the volume of data to be stored, it is possible to store (for example, in two dedicated numerical arrays referred to here as maps), for each position value (xo, yo) defining the exploration trajectory, the current maximum of the test statistic and the associated velocity index. At the end of the iteration, analysis of the maps can determine the detections and their associated velocity parameters.

[0187] It is important to note that the exploration trajectory, applied a posteriori to the image frame stack and not during acquisition as the pointing trajectory does, affects the movement of the moving object from one frame to the next (inter-frame effect) but, unlike the pointing trajectory, cannot affect the effect of the object's movement during the integration of each frame (intra-frame effect). Thus, for Figure 2(b), the pseudo-long-exposure image associated with tracking the SAR (12' stack) and summation along 17' will produce a smaller SAR image spot, and therefore one positioned with greater precision, than the pseudo-long-exposure image associated with sidereal tracking (12' stack) and summation along 17'. The average pointing trajectory (Figure 2(c)) is proposed to simultaneously minimize the extent due to the intra-frame movement of stars and SARs.The inter-frame extension is corrected by the exploration trajectory specific to each object.

[0188] It is possible to fail to detect moving space objects in some frame stacks. For example, several stacks may be recorded for different successive directions, where only some of the stacks contain moving space objects. For others, the images may have been taken in a region of the sky where no moving space objects are present, or sufficiently visible.

[0189] The dated frame coordinates of moving space objects are estimated. The dated frame coordinates of the detected moving space objects are converted into coordinates on the celestial sphere using the previously established coordinate relation.

[0190] The coordinates of a moving spatial object include its position.

[0191] The moving spatial object can also be characterized by its speed in the observed plane, and / or its average magnitude and / or its photometric variation.

[0192] According to one embodiment, the determination of the coordinates of a detected spatial object can be carried out according to one or more of the following variants.

[0193] According to a first variant, the position of the spatial object is approximated by its integer coordinates (x, y) associated with the maxima of the statistical test map calculated during the detection step.

[0194] According to a second variant, sub-pixel coordinates are obtained by a continuous estimator (such as the position of the center of gravity or interpolation in the vicinity of a vertex assumed to be locally parabolic) applied around the maximum, corresponding to the object detection, of an image map. The image map can be the statistical test performed or the pseudo-long-exposure image resulting from the exploration trajectory associated with the detection.

[0195] Advantageously, according to a third variant, the non-integer values ​​of the coordinates (a, b) and the velocities (u n , v nThe values ​​are sought by maximizing the signal-to-noise ratio (SNR). This simply requires using a classic iterative optimizer, familiar to those skilled in the art, applied to the calculation of the SNR defined below. This procedure can be initialized with integer or non-integer values ​​obtained by one of the two preceding variants. Advantageously, the calculation can be performed on only a sub-part in (x, y) of the image frame stack, extracted from the vicinity of the exploration path that led to the detection.

[0196] The coordinates on the celestial sphere thus measured are transmitted to a control unit or another computing unit, for example, for orbitographic calculations. Figure 4 is a schematic representation of steps in the monitoring process, and in particular steps in the processing phase including the detection of space objects, according to a non-limiting embodiment of the invention.

[0197] Figure 4 shows the two extreme frames (Æ = 0 and k = X-1) of a stack of K images, in which a space object has moved during acquisition. In the example shown, the space object is a star. Of course, it could also be a solar stellar object.

[0198] The stack of recorded frames is I(x, y, k) where k denotes the frame index and (x, y) the two-dimensional coordinates of the pixels of each frame.

[0199] The coordinates on the pixel grid of the successive positions of the star, or of the displacement vectors, are not a priori integers.

[0200] To exploit the fact that the spatial object's motion is rectilinear and uniform, the processing relies on adapted filtering, which consists of projecting the stack onto a series of known trajectories. These trajectories are defined by a point, for example the coordinates (a, b) in the first frame (Æ = 0), and a two-dimensional velocity with components (u n , v n ), expressed in pixels per frame.

[0201] Velocities are indexed by an index n which can take many values. For example, for satellites in circular orbit between the upper part of the medium Earth Orbit (MEO) orbital domain moving at about 50 arcseconds / s and the geostationary Earth Orbit (GEO) with very low speed, a granularity of 0.5" / s in the velocity grid leads to a grid of 100x 100 or 10000 velocities to consider.

[0202] In particular, for satellite searches without preconceived notions, setting a maximum search altitude leads to a maximum apparent velocity. It is then possible to explore all velocities (u, v) within a disk whose radius is this maximum velocity and whose granularity is adapted to the desired accuracy, the available computing power, and the expected signal-to-noise ratio (SNR). The calculation, or processing, can be broken down into two steps.

[0203] First, on each frame k, we calculate the signal S and noise B contributions on each trajectory:

[0204] [Math]

[0205] [Math2] where h is a model of the shape of the image spot, which can cover several pixels as illustrated by disk 41 in Figure 4, and has a very rapid decay.

[0206] Of course, to limit the computational load, for each value of (a, b) the sum over (x, y) can be limited to the few pixels defined by the support of h, that is, the area where h is significantly non-zero. This is recognized in ku n and kv n the dx(k) and dy(k) shifts mentioned previously, which could have been applied to I and w, but which are here applied to h, unique and of small support.

[0207] The term w(x, y) is a weighting map of each pixel in the stack based on its noise characteristics:

[0208] [Math3] f 0 if defective pixel w 2 (x, y) = -i, where cron is the standard deviation of the image sensor read noise determined during its preliminary calibration, and / Vfond, determined during preprocessing, is the average sky background level. Since these noises are uncorrelated, their variances are additive. It is assumed that the spatiotemporal variation of the background is negligible along the pointing trajectory during image stack acquisition.

[0209] Secondly, the total contributions per frame are calculated on each trajectory:

[0210] [Math4]

[0211] [Math5]

[0212] [Math6]

[0213] The final statistical test T n a statistical test based on adapted filtering, projection of the current signal weighted by its noise onto the expected signal and normalized by the projection of the expected signal onto itself.

[0214] A satellite is considered detected when T n exceeds a typical value of 7. We then have in (a, b) the associated position in a reference frame and in (u n , v n ) an estimate of the speed. An estimate of the magnitude is provided by S. These quantities, in coordinates or pixel level, are then converted into stellar data (angles, angular velocities and magnitudes) by the correspondence established on the reference stars, that is, using the coordinate and magnitude relations established previously.

[0215] Any false detections can be rejected in later steps, for example during revisit steps or orbitographic calculations.

[0216] The two-step calculation described above allows for the simple elimination of "outliers," that is, aberrant values ​​originating, for example, from cosmic rays. To do this, for each n, the one or two highest values ​​of the S n ,k, as well as their fî n k associated, are not taken into account in the calculation of Sn and B n For example, when the calculation is based on K = 100 frames, the systematic loss of 2 frames has very little effect on the result.

[0217] The difficulty with this calculation is that the final result can be written as RSB n(a, b), is of dimension (number of pixels, number of speeds), which can quickly lead to very high memory usage. Furthermore, as outlined in Figure 4, several speeds (indicated by the arrows) on the search grid, different from the actual speed but close on the quantization grid, will detect a strong presence signal corresponding to the same object. To overcome these two problems, and as detailed above, it suffices during the speed loop (n) to store, in only two arrays indexed in (a, b), the maximum value of the SNR. n (a, b) and the value of the associated n to ultimately determine the probability of a satellite's presence and its speed (u n , v n ).

[0218] It is unlikely that two objects such as RSOs could intersect on the same reference pixel. The less bright one would then go undetected. If necessary, a second pair of arrays storing the secondary maximum, combined with a "speed proximity threshold," would solve this problem, and other pairs would allow for handling even less probable cases if needed.

[0219] Advantageously, the method according to the invention also includes a phase of fine determination of the coordinates of a moving spatial object detected in the previous step.

[0220] When using a ground-based monitoring system, atmospheric effects on the light beam passing through the atmosphere must be taken into account when determining the coordinates of a space object. Indeed, the main limitation of angular position measurements of stars or celestial objects (COs) by ground-based means stems from the atmosphere. Atmospheric refraction introduces a bias exceeding one arcminute for elevations below 45°. This correction is achieved by measuring the coordinates of the detected COs and converting them to celestial coordinates using the established coordinate relation. The effects of turbulence, i.e., high-frequency random motion, average over the estimated position over a long period. A long-exposure image of any near-point-like celestial object is approximately a Gaussian with an angular width on the order of one arcsecond at the zenith in a good observing site.

[0221] For measuring the position of an image spot, the accuracy ct is on the order of ct = U / RS.SB, where L is the full width at half maximum (FWHM) of the spot and RSB is the signal-to-noise ratio. Thus, from spots limited by atmospheric turbulence in long exposures (width on the order of L = 2" for observations at 45° elevation in an average site), an accuracy on the order of 0.05" can be obtained with an RSB of 40.

[0222] With short-exposure images, a large part of the turbulence is eliminated. It is known that 90% of the effect of turbulence is contained in the fluctuation of the beam's angle of arrival. Moreover, most of this directional perturbation is at low frequencies, with a cutoff frequency of approximately 0.3v / λ, where v is the wind speed and D is the telescope diameter, or about 10 Hz for typical conditions (v = 10 m / s and D = 0.3 m). It is therefore known that, on a small telescope, once this fluctuation in the angle of arrival is eliminated within each short-exposure image, the effect of turbulence is reduced to the random movement of a smaller spot. An exposure time of 0.1 s thus reduces the extent L of each short-exposure image, and therefore the noise in its position measurement.Summation over several seconds, on the RSOs but also on the stars in the field, allows the position fluctuations between the frames to be averaged and makes the average tend towards the position identical to that obtained on the equivalent long-exposure image.

[0223] The precise determination of the coordinates can be achieved by one or more of the following steps or actions to take into account atmospheric effects.

[0224] The effects of variation in the field of atmospheric refraction, diurnal aberration or tube distortion can be modeled by using a polynomial of at least third order to establish the coordinate relationship.

[0225] Chromatic effects (due to atmospheric refraction or aberrations of a tube or lens containing dioptric elements) can be modeled by incorporating a correction term based on the spectral characteristics of the space object, known from a catalog, approximated, or measured by spectroscopy. The parallax effect induced by atmospheric refraction, which depends on the distance of the moving space object from the optical tube, can also be modeled.

[0226] For terrestrial or space-based surveillance systems, the precise determination of coordinates can still be achieved by one or more of the following steps or actions.

[0227] The following can be modeled:

[0228] - the movement of the moving space object, such as a satellite, and of the observer during the time it takes for light to travel between the satellite and the camera,

[0229] - the effect of annual aberration, experienced by stars but not by moving space objects, and / or

[0230] - in the case of using a scanning sensor, or rolling shutter sensor, the effects of variation in the field of the actual moment of measurement.

[0231] The steps or actions for the precise determination of coordinates can be applied to both stars and moving orbital objects. Some steps are reserved for RSOs (Remotely Seismic Organizations).

[0232] Figure 5 illustrates parasitic effects induced during the passage of light through the atmosphere. The density of the atmosphere, continuously varying between the ground 51 and space, is represented by two homogeneous layers 52, 53.

[0233] In Figure 5, the atmospheric effects considered below are drawn in an exaggerated manner for better visibility. The description first focuses on the main ray 54, shown in a thick solid line, and then on the ray 54' of another color, shown in a thick dashed line.

[0234] For a star 55 (very distant, appearing to come from a direction 0) observed by a T-shaped telescope, the pressure gradient, and therefore the refractive index, of the air bends the light rays, causing the star to appear after passing through the atmosphere as coming from the direction 1. This atmospheric refraction is characterized by an angle R, which is greater the greater the distance to the zenith, in the direction Z and characterized by the zenith angle z.

[0235] Thus, the compensation of the effect of atmospheric refraction is achieved by the coordinate relation, allowing the moving space object to benefit from the very good knowledge of the position of the reference stars present in the field.

[0236] Since the refractive index of air is chromatic, meaning it depends on the frequency of the radiation in question, the deviation R is also chromatic. The image of a broad-spectrum source, such as a star or a solar radiation system (SRS), appears as a small segment with an angular extent dR. The star's temperature, which determines the energy distribution among the frequencies detected by the sensor, then modifies the position of the apparent center of the segment and therefore the effective value of R.

[0237] Figure 5 shows that atmospheric deflection also induces a lateral displacement of amplitude b in the beam. This lateral displacement is not taken into account in astronomy because it has no impact for an object at infinity. If we consider a nearby object such as a 56 SAR (represented by a black disk), located at position 2 and therefore aligned with the star for an observer at position T, its apparent position will be perceived, depending on the color, at positions 1 and 1'. This deflection can reach several arcminutes. It is therefore important to apply to the SAR the angular correction R determined on the star. Applying this correction is equivalent to placing the SAR at position 0. This corresponds to making a parallax error relative to the actual position 2. The parallax, expressed as P = θ / h, is greater the lower the altitude h of the SAR.

[0238] It is therefore advantageous to include in the processing the deterministic numerical correction (once the distance h is known) of these terms P and dR (typically less than one arcsecond) after having applied the necessary parallax correction R (typically one arcminute) determined on the stars.

[0239] For the precise determination of the coordinates of a previously detected moving space object, the image stack is limited to a local sub-tube in the vicinity of the previously detected object. The parameters obtained by this precise determination include the position of the moving space object and an estimate of the uncertainty in the estimated position.

[0240] The parameters returned by the fine determination allow in particular to adjust the shape of an image spot in a stack of images, this spot being assumed to be of Gaussian shape beforehand for the detection of moving spatial objects.

[0241] According to some embodiments, the process according to the invention includes preliminary preparation or calibration steps.

[0242] In particular, the method according to the invention may include a preliminary step of determining the terrestrial coordinates of the telescope in the case of a terrestrial surveillance system.

[0243] This position can be refined by observing reference satellites and searching for telescope coordinates that minimize the gap between the measured dated positions of these satellites and the dated positions tabulated elsewhere.

[0244] Reference satellite observations can also be repeated regularly to ensure the quality of the measurements.

[0245] Observing space in "short-exposure image stacks" mode offers many technical advantages compared to observing from a single long-exposure image, depending on the operations performed.

[0246] It is possible to determine the direction of movement of objects.

[0247] The background level added to the signal from moving spatial objects is reduced, particularly in the infrared range, where the background level is very high. Indeed, as soon as a moving object leaves a pixel, continuing integration no longer provides a signal but only background noise, and its associated shot noise.

[0248] The average background level can be estimated locally on each pixel (for example, by a median filter applied according to k on the stack). Incidental events can be filtered out. Indeed, the loss of a few pixels in a long frame stack is less detrimental than in a single image.

[0249] This mode of observation also has the advantage of being able to accept a drift or a rotation (locally linear) around the line of sight, whether intentional (sliding acquisition) or suffered (poorly controlled pointing of a terrestrial or space telescope, field rotation induced by a non-equatorial mount).

[0250] The diameter of the image spots of observed SSRs, as well as those of reference stars, can be reduced, thereby increasing positioning accuracy. This is achieved in particular by filtering out much of the atmospheric turbulence, such as the low-frequency fluctuation of the arrival angle.

[0251] Image sensor sampling by spot movement is refined. Consequently, the estimation of a spatial object's position is also improved, whereas a single elongated image leads to a less accurate estimate. This refined sampling allows us to move beyond the classic "sidereal tracking" mode and benefit from the "sliding sensor" effect, even on stars. The "sliding sensor" effect, induced by the inherent movement of moving objects or, in the case of stars, by acquisition along a pointing trajectory different from the sidereal motion, leads to a beneficial averaging of local sensor inhomogeneities. In particular, it becomes possible to use, without significant degradation, sensors with a few defective pixels, which are less expensive than higher-quality sensors, especially for large dimensions.

[0252] Of course, the invention is not limited to the examples just described and many modifications can be made to these examples without departing from the scope of the invention.

Claims

36 DEMANDS 1. Method for monitoring moving space objects, the method being implemented by a ground-based or space-based monitoring system (100) comprising at least one optical tube or lens, at least one camera, a computing unit (11, 13) and a control unit (19), the method comprising the following steps: - pointing the tube or optical lens along a known trajectory on the celestial sphere, - sliding acquisition, by the camera, of at least one stack (12, 12') of image frames of space according to a field of view produced by the optical tube, each stack (12, 12') of image frames comprising at least three image frames, each image frame being recorded with an exposure time of less than 1 second, - detection, in the field of view, of a plurality of stars (14, 55), preferably several dozen, in at least one stack (12, 12') of image frames, - estimation of the coordinates, in at least one frame and at a given time, of the stars detected in the field of view, - identification of detected stars using a list of reference stars, previously extracted from a stellar catalogue, the list comprising several thousand reference stars, - establishment of a relationship, called a coordinate relationship, linking the celestial coordinates to the coordinates in the image frames of the stars present in the reference star list, - detection of at least one moving spatial object (15, 56) in at least one image frame stack and estimation of their dated frame coordinates, - conversion of the dated frame coordinates of each detected moving spatial object into coordinates on the celestial sphere using the coordinate relation, and - transmission of coordinates on the celestial vault to the control unit (19) or to another computing unit. 37 2. Method according to claim 1, further comprising the parameterized modeling of an image spot of a spatial object in the field of view, the image spot having a defined shape and movement, the spatial object comprising a star (14, 55) or a moving spatial object (15, 56).

3. A method according to any one of the preceding claims, characterized in that the step of detecting at least one space object comprises the following steps, the space object comprising a star or a moving space object: - choice of at least one exploration trajectory of the stack (12, 12') of image frames, the exploration trajectory being defined as a shift dx(k), dy(k) applied to each pixel (x, y) in the frame of index k, - calculation in the (12, 12') frame stack, along each exploration path, of a test statistic map for a plurality of pixels (x, y) and dx(k), dy(k) offsets of interest, and - possible extraction of spatial objects whose test statistic is greater than a threshold value.

4. A method according to the preceding claim, wherein the exploration trajectory corresponds to a uniform linear motion, resulting from the combination of the motion of the space objects and the pointing trajectory, the motion of the space objects and the pointing trajectory being substantially linear over the duration of the frame stack, the space objects comprising stars (14, 55) and / or moving space objects (15, 56).

5. A method according to any one of the preceding claims, characterized in that it further comprises a step of finely determining the coordinates of a detected space object, the space object comprising a star (14, 55) or a moving space object (15, 56), the finely determining comprising at least one of the following steps: - modeling the displacement of the moving space object and the observer during the light propagation time between the moving space object and the camera, - modeling of the effect of annual aberration, experienced by stars but not by moving space objects, - modeling, for a scanning sensor, of the effects of variation in the field at the effective time of measurement; - modeling the effects of variations in the field of atmospheric refraction, diurnal aberration, or optical tube distortion by using a polynomial of at least third order to establish the coordinate relation, - Modeling of chromatic effects by taking into account a correction term based on the spectral distribution of the spatial object, - modeling of the parallax effect induced by atmospheric refraction.

6. A method according to any one of the preceding claims, wherein the detection of at least one moving space object (15, 56) includes determining its velocity in the observed plane, its average magnitude and / or its photometric variation.

7. A method according to the preceding claim, characterized in that the list of reference stars is determined from a star catalogue by: - the selection of stars based on at least one of the following characteristics: magnitude, temperature, multiplicity, variability, distance to neighboring stars, and / or - updating the coordinates of the stars in the list with the Earth's orbital parameters and the characteristics of the stars, at a date close to the observation.

8. A method according to any one of the preceding claims, characterized in that it further comprises a preprocessing step of each stack (12, 12') of image frames, prior to the step of detecting a plurality of stars (14, 55) in the field of view.

9. A process according to the preceding claim, characterized in that the pretreatment step comprises one or more of the following steps: - Filtering of frames in the (12, 12') frame stack to eliminate a static background, - pixel grouping of the frames, - reduction of the field of view, and / or - correction of frames on the calibration database of the monitoring system.

10. A method according to any one of the preceding claims, characterized in that the pointing trajectory corresponds to - a stellar chase, - a zero pointing trajectory in which the optical tube is in fixed aiming, or - a pointing trajectory with average speed weighted between the apparent speeds of the stars and the expected moving objects.

11. A method according to any one of the preceding claims, characterized in that it further comprises fixing the camera integration time for each frame so that during the acquisition time, the movement of a spatial object covers less than 10 pixels of one frame of the stack (12, 12') of frames, and preferably less than one pixel.

12. A method according to any one of the preceding claims, further comprising precise timing, on the order of milliseconds, of a frame of the stack (12, 12') of frames, preferably at a central instant of the stack.

13. A method according to any one of the preceding claims, characterized in that it further comprises a preliminary step of determining the terrestrial coordinates of the surveillance system (100).

14. A method according to any one of the preceding claims, characterized in that it further comprises a preliminary or regular calibration step of the monitoring system (100) by observation of at least one reference satellite.

15. System (100) for monitoring moving space objects, comprising: - at least one optical tube or lens, configured to be pointed along a known trajectory on the celestial sphere, - at least one camera, configured for the sliding or fixed acquisition of a stack of frames of space according to a field of view produced by the tube or optical lens, the frame stack comprising at least three frames, each frame being recorded with an exposure time of less than 1 second, - a computing unit (11, 13), - a control unit (19), the system (100) being configured to implement a method for monitoring moving space objects according to any one of the preceding claims.

Images