Hyperspectral imaging optical system

Abstract

Hyperspectral imaging optical system (1) comprising: an optical system called a telescope (T) adapted to collect light (L) from a scene to be observed moving along a direction A so as to form an image called an intermediate image in an image focal plane of said telescope, at least a hyperspectral detection assembly (ED) of at least a portion of the light (L) collected by said telescope and comprising: a filter assembly (EF) comprising upstream spectral filters arranged near the image focal plane of said telescope, each upstream spectral filter being bandpass to an upstream spectral band distinct from the others and being arranged so as to spectrally filter a respective sub-beam associated with a field of view (FOVi) of the observed scene, an optical system called an imaging lens (OI),an object focal plane (of said imaging lens being coincident with the image focal plane of said telescope), a matrix detector (Det) adapted to acquire an image of the intermediate image by the imaging lens, the imaging lens being configured to form spectrally distinct sub-images corresponding to said fields of view of the scene associated with the sub-beams ( ), each of the sub-images being formed on a sub-region ( ) of the detector distinct from the others.

Description

Title of the invention: Hyperspectral imaging optical system technical field

[0001] The present invention relates to the field of hyperspectral imaging and in particular to pushbroom (or linear-field) type hyperspectral imagers. Prior art

[0002] A hyperspectral imaging system allows the acquisition of a three-dimensional image of a scene to be observed: two classical spatial dimensions and a spectral dimension that corresponds to the decomposition in the spectral domain of each point of the imaged object. For clarification, a "hyperspectral imaging system" is understood here to be an imaging system capable of simultaneously imaging at least 60 distinct spectral bands. In contrast, a "multispectral imaging system" typically images ten or fewer spectral bands simultaneously.

[0003] There are several types of hyperspectral imaging systems that differ in the method of acquiring the different points of the scene to be observed (point by point, line by line or full field) and in the method used to decompose the light collected by the imaging system according to several wavelengths (diffraction, refraction, interferometry...).

[0004] Pushbroom imaging systems are the most commonly used in remote sensing and airborne systems. They are typically used in situations where there is relative movement between a scene to be observed and the imaging device. This movement may be due to the device being mounted on a vehicle, aircraft, or satellite moving over an area of ​​interest, with the device oriented so that the area of ​​interest intersects the device's field of view. Alternatively, the movement is achieved with a scanning mirror. Alternatively, the device may be stationary and the area of ​​interest may move across the field of view, for example, on a conveyor belt.The hyperspectral image acquired by a pushbroom imager is generally a band image obtained by imaging a single band (linear field) of the scene to be observed in the field of view, the relative displacement allowing successive acquisition of a plurality of continuous bands.

[0005] Figure IA schematically illustrates a prior art pushbroom hyperspectral imager P comprising a telescope optical system T and a spectro-imager SP. Figure IB illustrates in more detail the various elements of a known prior art spectro-imager SP. Typically, the spectro-imager SP comprises a slit F placed in the image focal plane PF of the telescope T which collects the light L from The scene to be observed (Obj.) The telescope T is a known optical system adapted to produce the image of an object located at infinity relative to the telescope's focal length. It is known to consist of a plurality of off-axis parabolas and / or mirrors or lenses.

[0006] In the example illustrated in [Fig. 1A], the relative displacement of the scene with respect to the imager P occurs along direction A, perpendicular to the direction of the slit F, called direction B. Thus, only a limited portion FOV of the field of view of the scene to be observed, called the linear field, passes through the slit F to be imaged on the detector Det of the spectro-imager SP. The image of the portion of the scene observed through the slit is called the linear image.

[0007] The spectro-imager SP includes an optical collimation system Col adapted to collimate the LF portion of the light passing through the slit F. The optical collimation system is arranged so that its object focal plane PFo,coi coincides with the plane containing the slit F and the image focal plane P Fit of the telescope T. Alternatively, spectro-imagers may use concentric optical setups of the Dyson or Offner type in which the beam is not collimated at the level of the dispersive element.

[0008] A dispersive optical element Disp is adapted to spread the spectral information by deflecting the rays according to their wavelength. This dispersive optical element can be a prism, a diffraction grating, or a combination of disjoint or grouped transmission dispersive optical elements. The dispersive element has a dispersion axis along the direction A. Thus, after passing through the dispersive element, the angle of deviation of the rays in the LF portion reflects their wavelength. An imaging optical system O located downstream of the telescope along the optical path is adapted to spatially separate the different wavelengths and focus the LF portion that has passed through the dispersive optical element onto a matrix detector Det placed in the image focal plane of the imaging optical system O. The image focal plane of the telescope containing the slit F is therefore the conjugate of the image focal plane of the imaging optical system O containing the matrix detector.Thus, the optical imaging system O makes it possible to produce a plurality of imagelets of the slit F corresponding to the restricted portion of the field of view FOV of the scene to be observed, each associated with a distinct spectral band.

[0009] Without a slit F in the object focal plane of the collimator, the beam being angularly extended in the direction of dispersion, after passing through the dispersive optical element, there would be spatial overlap on the detector Det of the different wavelengths associated with different points of the scene. It would then be impossible to spatially translate, on the matrix detector, the hyperspectral information of the imaged scene. The slit F imaged on the detector Det therefore allows a spatial separation of the wavelengths on the detector by imaging an entire line of the scene to be observed Obj.

[0010] The matrix detector is typically arranged so that each image is oriented along an axis c (“column”) of the detector's pixel matrix, the imaging optical system being adapted so that the image images are aligned along an axis A (“row”), perpendicular to the axis. Thus, in the absence of distortion, each column of the detector is exposed to radiation corresponding to the same linear field of view but at different wavelengths. Each row of the detector records the spectrum of each point in the imaged field of view.

[0011] Hyperspectral instruments known to those skilled in the art and based on dispersive systems have constraints regarding signal-to-noise ratio, flexibility of the chosen bands, and stray light. For example, it is not possible to optimize the signal-to-noise ratio of a particular spectral band relative to another because all spectral bands are separated and detected in the same way.

[0012] The invention aims to overcome certain problems of the prior art by developing a hyperspectral imaging optical system comprising more than 60 spectral filters arranged in the image focal plane of the telescope observing a moving scene. Thus, the system of the invention, through the modularity of the filters, allows for a very good signal-to-noise ratio in all spectral bands, complete flexibility in the bands selected, and controlled stray light. Summary of the invention

[0013] To this end, an object of the invention is a hyperspectral imaging optical system comprising: - an optical system called a telescope adapted to collect light from a scene to be observed moving along a direction A so as to form an image called an intermediate image in a focal plane image of said telescope, - at least one hyperspectral detection array of at least a portion of the light collected by said telescope and comprising: - a filter array comprising n > 60 upstream spectral filters arranged near the image focal plane of said telescope, each upstream spectral filter being bandpass for a distinct upstream spectral band and arranged to spectrally filter a respective sub-beam associated with a field of view of the observed scene, - an optical system called an imaging lens, an object focal plane of said imaging lens coinciding with the image focal plane of said telescope - a matrix detector adapted to acquire an image of the intermediate image via the imaging lens, the imaging objective being configured to form n spectrally distinct sub-images corresponding to the n fields of view of the scene associated with the n sub-beams, each of the n sub-images being formed on a sub-region of the detector distinct from the others.

[0014]

[0015]

[0016]

[0017]

[0018]

[0019]

[0020]

[0021]

[0022]

[0023] According to a particular embodiment, the upstream spectral filters and said subregions extend along a principal direction substantially perpendicular to an optical axis of the imaging lens and substantially perpendicular to said direction A. According to a particular embodiment, a so-called lateral dimension of each upstream spectral filter along direction A is adapted so as to obtain a signal-to-noise ratio greater than a predetermined value in the sub-image associated with said filter upstream spectral. According to a particular embodiment, each of the sub-regions comprises a number of pixels along direction A determined as a function of, on the one hand, a large dissement of said imaging objective and on the other hand of a dimension called lateral along the direction A of the upstream spectral filter filtering said spectral band. According to a particular embodiment, the telescope has a number of apertures greater than 5, preferably greater than 7, and even more preferably greater than 10. Preferably, a so-called lateral dimension of each upstream spectral filter along direction A is greater than 50, preferably greater than 80 / mi. According to a particular embodiment, the magnification of the imaging lens G is determined based on the predetermined GSD ground resolution according to the following formula GSD _ Hxp, with H being the distance between the scene and the system, the ~ f^GT focal length image of the telescope T, P the pixel pitch of the detector along the direction A. According to a particular embodiment, the imaging lens has a magnification of less than 1 in order to obtain a ground resolution of approximately 20 m for said system. According to a particular embodiment, the upstream spectral filters form a single-piece structure, the upstream spectral filters being glued together. According to a particular embodiment, the upstream spectral filters are separated by a distance of less than 25 [tm, preferably less than 10 fim. According to a particular embodiment, the upstream spectral filters are added or attached to a transparent plate exposed to said light. According to a particular embodiment, the system comprises n >100 upstream spectral filters, preferably n > 200 upstream spectral filters so that said filter set transmits a spectral band called the total spectral band from 400 nm to 2500 nm.

[0024] According to a particular embodiment, the hyperspectral detection set further comprises n downstream spectral filters located near the detector and bandpass to a downstream spectral band distinct from the others, a downstream spectral filter being arranged so as to filter a subbeam associated with an upstream spectral filter, a downstream spectral band of a downstream spectral filter having a non-zero overlap with the upstream spectral band of the associated upstream spectral filter.

[0025] According to a particular embodiment, the system comprises a plurality w > 1 of hyperspectral detection sets each adapted to transmit and detect a spectral band called the total spectral band different from the others, said system comprising a dichroic splitter arranged on the optical path of the light to spatially separate said light collected by the telescope into m beams having different spectral ranges and to direct each of said m beams to a hyperspectral detection set adapted to detect said spectral range of said beam. Brief description of the drawings

[0026] Other features, details and advantages of the invention will become apparent from the description given with reference to the accompanying drawings provided by way of example, which represent, respectively:

[0027] [Fig.1A] and [Fig.1B] a schematic view of a prior art pushbroom type hys-perspectral imaging optical system;

[0028] [Fig.2A], [Fig.2B] and [Fig.2C] a schematic view of an optical imaging system hyperspectral of the invention;

[0029] [Fig.3] a schematic view of the upstream spectral filters of a hyperspectral imaging optical system of the invention;

[0030] [Fig.4] and [Fig.5] a schematic view of respectively an embodiment of upstream spectral filters of the hyperspectral imaging optical system of the invention.

[0031] [Fig.6] and [Fig.7] a schematic view of a hyperspectral imaging optical system according to embodiments of the invention.

[0032] In the figures, unless otherwise indicated, the elements are not to scale. Description of the implementation methods

[0033] Figures 2A and 2B present a schematic view of a pushbroom-type hyperspectral imaging optical system 1 according to the invention. Figure 2A illustrates the arrangement of the system 1 with respect to the scene to be observed Obj. Figure 2B is a more detailed schematic representation of the various elements of the system 1 along a plane 31.

[0034] As in prior art pushbroom devices, system 1 comprises an optical system called a telescope T adapted to collect light L from the scene at Observe Obj. The telescope T is a catadioptric system known in itself, adapted to produce the image of an object located at infinity relative to the focal length of the system in the image focal plane PF im T of the telescope. As is known, it can consist of a plurality of off-axis parabolas and / or mirrors or lenses. As an illustrative example, in the embodiment of [Fig. 2B], the telescope comprises two off-axis parabolas and one mirror.

[0035] To acquire a hyperspectral image of the scene Obj, the system 1 includes a hyperspectral detection ED set of at least a portion of the light L collected by the telescope of the system 1. Figure 2C is a schematic representation along the Ax plane which details more precisely the hyperspectral detection ED set which includes in particular an imaging lens 01 and a matrix detector Det.

[0036] The image of the scene formed by the telescope is called the intermediate image F because the hyperspectral detection assembly ED of system 1 includes an imaging lens OI with an object focal plane PFo>Oi coinciding with the image focal plane PFim j of the telescope. The imaging lens OI produces the image of this intermediate image It on the detector Det. The matrix detector Det is a CCD sensor, a CMOS sensor, or any matrix detector known from the prior art.

[0037] By way of illustration, in the embodiment shown in Figures 2B and 2C, the imaging lens comprises two distinct optical groups: a first group 01 producing the image of the intermediate image, referred to as the first image, at infinity, and a second group 02 producing the image of the first image in the image focal plane of the second group PF^oi. Alternatively, according to another embodiment, the imaging lens comprises a different number of lens groups.

[0038] In the example illustrated in Figure 2A, the relative motion of the scene to be observed with respect to system 1 is along direction A. For example, system 1 may be mounted on a vehicle or aircraft moving such that the fixed scene to be observed, Obj, intercepts the field of view (FOV) of the telescope T. Alternatively, system 1 may be fixed and the scene to be observed, Obj, may move across the field of view, for example, on a conveyor belt. Hereafter, the terms "direction A" and "direction of movement" are considered equivalent.

[0039] System 1 of the invention differs from prior art hyperspectral pushbroom devices at least in that the ED assembly comprises a filter assembly EF enabling hyperspectral operation of System 1. More specifically, the filter assembly EF comprises n > 60 upstream spectral filters Ri-.. R^ arranged near the image focal plane PF im / of the telescope. Preferably, as illustrated in Figures 2B and 2C, the upstream spectral filters are arranged in the image focal plane PF of the telescope in order to reduce the vignetting produced by each filter. Unlike prior art multispectral devices comprising dispersive elements (see figures 1A-1B), system 1 of the invention does not include a slit F in the image focal plane of the telescope.

[0040] Each upstream spectral filter Rj is bandpass for an upstream spectral band distinct from the others and is arranged to spectrally filter a respective subbeam SFj associated with a field of view (FOV) of the observed scene. Thus, the subbeam SFj presents spectral information only in the spectral band A4. Given the optical design of system 1, the imaging lens 01 is adapted to form n spectrally distinct sub-images, corresponding respectively to the n fields of view (FOV) of the scene associated with the n subbeams SFj. Each of the n sub-images is formed on a sub-region P of the detector distinct from the others. In each sub-image, only a restricted portion of the scene is observed (corresponding to the field of view (FOV) of each sub-image). The set of fields of view (FOV) associated with each subbeam SFj corresponds to the field of view (FOV) of the telescope.The size of each sub-image I depends on the size of the associated filter Ri and the magnification of the imaging lens. The size of each sub-region Pi along the A direction is adapted according to the size of the associated sub-image. As is known, this size can also be adapted according to the time averaging required to adjust the signal-to-noise ratio of the spectral band associated with the sub-region Pi (see below). Furthermore, according to the invention, as will be explained later, the size of the filters Rj along the A direction (and the size of the associated sub-regions Pi along the A direction) is not necessarily identical for each filter. Indeed, these dimensions can be different to optimize the signal-to-noise ratio (SNR) in the different A4,- bands.

[0041] Given the small field of view imaged in the sub-images, it is necessary to acquire several of them in order to reconstruct a relevant hyperspectral image of the scene Obj. Therefore, it is necessary to acquire a plurality p > 1 of sub-image sets I(q) = (q) ... Iₙ(q), each of the sub-image sets 7_j cor corresponding to a portion of the scene different from the others and imaged by the system at indexed instants qs [1, p] different from the others during the scrolling of the scene along direction A. The different sub-images are then juxtaposed to create so-called final images 7f> / = (7.(1), 7,(^), ..., I^p)}

[0042] In Figure 2C, to facilitate understanding of the scheme, three upstream spectral filters R ], Rp RN (and the fields of view FO FOVj, FOVN& are shown the 5Fh SFp SF N sub-beams respectively associated) although the device of the invention comprises more than 60 upstream spectral filters.

[0043] The use of n > 60 filters in the invention, rather than a dispersive element as in prior art devices, offers numerous advantages thanks to the modularity afforded by the filters. In particular, the filters allow for optimization of the signal-to-noise ratio (SNR) in all upstream spectral bands separately and enable complete flexibility in the selection of upstream spectral bands (see below). Conversely, the use of a dispersive element requires the use of a single matrix detector across the entire spectrum because each line of the detector records the spectrum of each point in the imaged field of view.

[0044] Furthermore, the filters of the invention make it possible to reduce parasitic noise. Indeed, performing spectral filtering in the image focal plane of the telescope significantly reduces the noise associated with diffuse reflections caused by passing through the imaging lens and detected by the detector, because each sub-beam is spectrally filtered before passing through the imaging lens. This differs in particular from the arrangement implemented in certain prior art multispectral devices in which a small number (typically less than 10) of spectral filters are located near the matrix detector.

[0045] Preferably, the EF filter set comprises n > 100 upstream spectral filters, preferably n > 200 upstream spectral filters, such that the EF filter set transmits a spectral band, referred to as the total spectral band, ranging from 400 tvn to 2500 nm. Thus, system 1 of the invention makes it possible to perform hyperspectral imaging covering the visible range up to the SWIR (Short Wavelength Infrared) range.

[0046] In hyperspectral devices, telescopes have a low aperture number (V) – typically less than 3 – to maximize the signal-to-noise ratio of the sub-images. A low aperture number allows for the collection of a greater flux. For reference, the aperture number NT of the telescope T is NT = f dT, where fT is the image focal length of the telescope and dT is the diameter of the entrance pupil of the telescope T. After numerous tests and simulations, the inventors realized that, in the invention, using n > 60 filters near the PFim,r plane with a telescope having an aperture number less than 5 produced a beam in the image focal plane that was too large transversely and induced excessive vignetting in the sub-images.To overcome this problem, the inventors have determined a preferred embodiment, denoted MP, in which the telescope T of the invention has an aperture number A∞ greater than 5, preferably greater than 7 and even more preferably greater than 10.

[0047] Thus, the collected light L is focused by the telescope T into a sufficiently narrow beam FF in the image focal plane PFimj to guarantee performance

[0048]

[0049]

[0050]

[0051]

[0052]

[0053] optimal filter placement. This allows, for example, limiting crosstalk between filters. The use of a telescope T with a high aperture number, as in the invention, is particularly counterintuitive because it reduces the flux collected by the telescope and therefore leads to a decrease in the signal-to-noise ratio (SNR). However, the invention makes it possible to compensate for this reduction in flux through time averaging or by using filters with appropriately sized lateral dimensions (see below), which was not possible in prior art devices using dispersive elements. Preferably, in the MP embodiment, the lateral dimension hf of each upstream spectral filter Rj is greater than 50 Rm, preferably greater than 80 pm, and even more preferably greater than 90 pm, in order to reduce vignetting of the FF beam by the upstream spectral filters. Even more preferably, the lateral dimension h of each upstream spectral filter R is between 90 Rni and 750 pm, this lateral dimension being preferably chosen according to the power spectral density of the light L in the upstream spectral band filtered by the upstream spectral filter Rf. Indeed, it is possible to compensate for a low power spectral density in the upstream spectral band by a relatively larger lateral dimension h (see below). Preferably, the imaging lens has a magnification G adapted according to the telescope's aperture number T in order to obtain a predetermined ground resolution in the system of the invention. Indeed, in the device of the invention, it is shown that the ground resolution (also called GSD for Ground Sampling Distance) is given by the following formula: GSD = with H the distance between the scene and system 1 (see figure 2A), f T the image focal length of telescope T, P the pixel pitch of the detector along direction A and G the magnification of the imaging lens. For example, according to one embodiment MP, the magnification G of the imaging lens is less than 1 to compensate for the relatively high aperture number Nt in order to obtain a predetermined ground resolution in the system of the invention. According to a preferred embodiment MP, the imaging lens has a magnification less than 1 in order to obtain a ground resolution of approximately 20 m in system 1. For example, for an aperture number Nt > 7 of the telescope T, the magnification G of the imaging lens is < 0.8. In one specific example, Nt = 9.6 and G = 0.73. The possibility of optimizing the GSD from the magnification of the relay optics is an option that was not possible in state-of-the-art devices using dispersive elements (see Figures IA and IB) in which the ground GSD resolution is given by the following relationship:

[0054] GSD =HxdFl fT

[0055] with H the distance between the scene and the device, fT the image focal length of the telescope T and dF the transverse dimension of the slit F (along the direction A).

[0056] Figure 3 schematically illustrates only the filter assembly EF and the matrix detector Det and shows how matching the filters and the detector allows for matching the SNR associated with each upstream spectral band AXZ. As an illustrative example, in Figure 3, the upstream spectral filters are all arranged in the image focal plane PFimJ of the telescope. Furthermore, the upstream spectral filters are rectangular and extend along the principal direction 2, which is substantially perpendicular to the optical axis AOO1 of the imaging lens OI and substantially perpendicular to the direction A. Similarly, to match the detection, each subregion Pi extends along the principal direction. Alternatively, in a different embodiment from that illustrated in [Fig. 3], the upstream spectral filters are square, elliptical, or circular.

[0057] According to a first embodiment, this SNR optimization is performed via the lateral dimension ht of each upstream spectral filter Ri along the direction A. Indeed, in order to improve the signal-to-noise ratio in a given upstream spectral band A\, it is possible to increase the lateral dimension hj of the upstream spectral filter Ri transmitting this spectral band, as this allows the transmission of the sub-beam SFj with a higher photon flux in this spectral band. Also, in the first embodiment, the lateral dimension is adapted so as to obtain a signal-to-noise ratio greater than a predetermined value in the sub-image A associated with the filter R,. This embodiment is particularly advantageous when the spectral power density of the light L in an upstream spectral band A\ filtered by the upstream spectral filter R, is low.This may be the case, for example, when the device user requires a low spectral resolution (i.e., a low bandwidth AX), for example 10 nm, in a given spectral band for a specific use.

[0058] By way of non-limiting example, the filters R^... R„ transmit adjacent spectral bands AXj... AX„ with an increasing center frequency from filter R{ to filter Rn. Moreover, as illustrated in the example in Figure 3, the lateral dimension of the filters decreases from the first filter R\ to filter R7?„, for example, because the light L has a decreasing spectral power density from the AXj band to the AX„ band.

[0059] According to a second embodiment, the SNR improvement in the upstream spectral band AXZ is achieved by time averaging in the Pi subregion associated with AX; Indeed, in order to improve the signal-to-noise ratio in an AXZ band, it is possible to average along the A direction sub-images lic / Y, each obtained in the Pp sub-region at a different time. This temporal averaging of sub-images in the P region is made possible by the number of pixel lines along the A direction in the P region, and allows obtaining a sub-image with a better signal-to-noise ratio. Recall that the dimension of a sub-image I on the detector depends on the lateral dimension h, the associated filter, and the magnification of the imaging lens 01. Also, according to the second embodiment, to optimize the signal-to-noise ratio in the upstream spectral band AX;S, the number of pixels along the A direction of the Pi sub-region associated with this band depends on the magnification of the imaging lens 01 and on the lateral dimension A; of the upstream spectral filter Ri transmitting the AX;S band.Like the first embodiment, the second embodiment improves the SNR, for example, when the power spectral density of light L in an upstream spectral band A transmitted by the upstream spectral filter Ri is low. It is understood that the number of lines along the direction may differ between the subregions R, depending on the specific characteristics of the associated bands AX.

[0060] By way of non-limiting example, in the example of Figure 3, the subregions Pn and Pi associated with the filters Rn and R, respectively, exhibit more lines along the A direction than the subregion P^ associated with the filter R\ in order to allow temporal averaging of the sub-images I,^q) and thus improve the SNR in the spectral bands AXn and AX,. This embodiment is advantageous when these bands AX1; and AX,- are more absorbed by the atmosphere than the band AX„, for example.

[0061] The first and second embodiments allow, through the use of filters, a high degree of modularity of the device 1 depending on the type of illumination and the imaged scene by optimizing the SNR on each of the spectral bands. This constitutes an advantage compared to prior art pushbroom-type hyperspectral devices using dispersive elements in which the detection of the different spectral bands could not be optimized separately.

[0062] Figure 4 illustrates an embodiment in which the upstream spectral filters form a single-piece structure. That is, the upstream regions R are stacked one on top of the other in direction A. The manufacturing process for these upstream spectral filters is thus simplified. The regions R can, for example, be attached or bonded together. Preferably, in the embodiment of Figure 4, the upstream spectral filters are separated by a distance of less than 25 µm, preferably less than 10 µm, which is due to the presence of adhesive between the filters.

[0063] Figure 5 illustrates an embodiment in which the upstream spectral filters are attached or bonded to a transparent LT plate to the light L collected by the telescope. Here, "transparent" refers to a transmission greater than 99%. The advantage of this method is that it allows the upstream spectral filters to be deposited directly onto the LT plate, thus fabricating the entire EF assembly in a single step.

[0064] Figure 6 schematically illustrates an embodiment of the invention, in which the hyperspectral detection ED assembly further comprises n downstream spectral filters ^2 / , located near the detector and bandpassing a downstream spectral band AT2( distinct from the others). Each downstream spectral filter Φ2i is arranged to filter a sub-beam SF{ associated with an upstream spectral filter Rj. Moreover, a downstream spectral band of a downstream spectral filter has a non-zero overlap with the upstream spectral band of the associated upstream spectral filter. Thus, the upstream spectral filter Ri and the associated downstream filter R^i are bandpassing a common spectral portion of the collected light. In this way, the sub-image is representative of the spectral information of the observed scene associated with this field of view and this common spectral portion.

[0065] Thanks to the use of upstream spectral filters combined with downstream spectral filters, the noise associated with diffuse reflections caused by passing through the imaging lens and detected by the detector is greatly reduced.

[0066] Furthermore, the intensity of the portion of light passing through the imaging lens and reflected by each downstream spectral filter R^i back to the imaging lens then exhibits a spectrum such that: [A / L - A42]• Depending on how the downstream and upstream spectral filters are chosen, the spectral bandwidth of the portion reflected by each downstream spectral filter can be very small or even zero if the associated upstream and downstream spectral bands are identical. Critically, even if the spectral bandwidth of the reflected portion is not zero, when the portion reflected by each downstream spectral filter is reflected by the different components of the imaging lens, it will be spectrally filtered by the other regions R# of the filter assembly EF and will therefore not be detected by the detector Det. Indeed, the R# regions are each bandpass with a downstream bandwidth distinct from the spectral band [A / h - A / lJ] of the reflected portion.Thanks to the upstream spectral filters, the contribution of these rays to stray light with respect to the sub-image associated with the R# region is greatly reduced.

[0067] Figure 7 schematically illustrates an embodiment of the invention in which the system 1 comprises a plurality 1 of hyperspectral detection assemblies ED, ED', each adapted to transmit and detect a so-called total spectral band different from the others. In the embodiment of Figure 7, the system comprises a dichroic beam splitter MD—for example, one or more dichroic mirrors—arranged along the optical path of the beam FF to spatially separate it into m beams exhibiting different spectral ranges and to direct each of the m beams to a hyperspectral detection array adapted to detect the spectral range of that beam. As an illustrative example, in the embodiment shown in [Fig. 7], system 1 comprises two paths separated by a dichroic blade or mirror MD and two hyperspectral detection arrays ED, ED'.

[0068] The embodiment of [Fig. 7] maximizes the detection efficiency of the detector Det, Det' of each ED, ED' hyperspectral detection assembly. Indeed, the detector of the hyperspectral detection assembly is specifically optimized for the spectral range of the beam directed into this assembly.

Claims

Demands

1. Hyperspectral imaging optical system (1) comprising: - an optical system called a telescope (T) having an aperture number greater than 5 and adapted to collect light (L) from a scene to be observed moving along a direction A so as to form an image called an intermediate image in an image focal plane (PFim^) of said telescope, - at least a hyperspectral detection assembly (ED) of at least a portion of the light (L) collected by said telescope and comprising: - a filter assembly (EF) comprising n > 60 upstream spectral filters (R^) arranged near the image focal plane (PFjmt) of said telescope, each upstream spectral filter being bandpassable to an upstream spectral band (Aà) distinct from the others and being arranged so as to spectrally filter a respective subbeam (SF,) associated with a field of view (FOVi) of the observed scene, - an optical system called an imaging lens (01),an object focal plane (PFoO^) of said imaging lens being coincident with the image focal plane of said telescope - a matrix detector (Det) adapted to acquire an image of the intermediate image by the imaging lens, the imaging lens being configured to form n spectrally distinct sub-images (1^) corresponding to said n fields of view (FOVj) of the scene associated with the n sub-beams (SF^), each of the n sub-images being formed on a sub-region (P^) of the detector distinct from the others.

2. A system according to the preceding claim, wherein said upstream spectral filters and said subregions extend along a principal direction substantially perpendicular to an optical axis (AOOI) of the imaging lens (01) and substantially perpendicular to said direction A

3. 2A. A system according to the preceding claim, wherein a dimension said lateral (h,) of each upstream spectral filter (Rj) along the direction A is adapted so as to obtain a signal-to-noise ratio greater than a predetermined value in the sub-image ( / .) associated with said upstream spectral filter.

4. System according to claim 2 or 3, wherein each of the subregions (Pi) comprises a number of pixels along the direction A determined as a function of, on the one hand, a magnification of said imaging lens (01) and on the other hand a so-called lateral dimension (h,) along the direction A of the upstream spectral filter filtering said spectral band.

5. A system according to any one of the preceding claims, wherein said telescope has an aperture number greater than 7, more preferably greater than 10.

6. System according to the preceding claim, wherein a so-called lateral dimension of each upstream spectral filter along the direction A is greater than 50, preferably greater than 80 ym.

7. System according to any one of the preceding claims, wherein the magnification of the imaging lens G is determined as a function of the predetermined ground resolution GSD according to the following formula: GSD = ^ with H the distance between the scene and the system, fT the image focal length of the telescope T, P the pixel pitch of the detector along the direction A.

8. System according to any one of the preceding claims, wherein said imaging lens has a magnification of less than 1 in order to obtain a ground resolution of about 20 m for said system.

9. System according to any one of the preceding claims, wherein the upstream spectral filters form a single-piece structure, the upstream spectral filters being glued together.

10. System according to the preceding claim, wherein the upstream spectral filters are separated by a distance of less than 25 finite, preferably less than 10 [im.

11. System according to any one of claims 1 to 8, wherein the upstream spectral filters are attached or bonded to a transparent plate (LT) to said light (L).

12. A system according to any one of the preceding claims, comprising n > 100 upstream spectral filters, preferably n > 200 upstream spectral filters such that said set of filter (EF) transmits a spectral band called the total spectral band ranging from 400 nm to 2500 nm.

13. System according to any one of the preceding claims, wherein the hyperspectral detection assembly further comprises 11 downstream spectral filters (Rzi), located close to the detector and bandpass to a downstream spectral band j distinct from the others, a downstream spectral filter being arranged to filter a subbeam (5F;) associated with an upstream spectral filter, a downstream spectral band (^2() of a downstream spectral filter having a non-zero overlap with the upstream spectral band ( A22 / ) of the associated upstream spectral filter.

14. A system according to any one of the preceding claims, comprising a plurality m > 1 of hyperspectral detection assemblies (ED, ED') each adapted to transmit and detect a spectral band called the total spectral band different from the others, said system comprising a dichroic splitter (MD) arranged on the optical path of the light to spatially separate said light collected by the telescope into ttl beams exhibiting different spectral ranges and to direct each of said m beams to a hyperspectral detection assembly adapted to detect said spectral range of said beam.

Images