On-board calibration of a pixel array of an infrared sensor

Abstract

The invention relates to a method for in situ calibration of a pixel array of an infrared sensor, said infrared sensor comprising: an optical system (2); and a pixel array (3) disposed in the focal plane (4) of the optical system (2), said optical system (2) allowing radiation to be focused on the pixel array (3). The method comprises the following steps carried out in a processing unit (7) connected to the infrared sensor, with the pixel array being characterised by a pixel gain table: acquisition (E2) by the pixel array of a first sequence of images of the radiation from a black screen (CN) corresponding to a scene at a first temperature; acquisition (E4) by the pixel array of a second sequence of images of the partially reflected radiation from the pixel array corresponding to a scene at a second temperature; processing (E5, E6, E7, E8, E9) of the first and second image sequences such as to update the gain table by calculating a correction representative of the aging and response of the pixels of the pixel array.

Description

[0001] DESCRIPTION

[0002] TITLE: On-board calibration of an infrared sensor pixel array

[0003] GENERAL TECHNICAL FIELD

[0004] The field of the invention is that of the preventive maintenance of matrix sensors used in infrared imaging devices. The invention relates more particularly to a method and a device for updating a calibration of such a matrix integrated in an infrared imaging device and more particularly, allowing a correction of disparities between the pixels of the matrix, without dismantling the matrix or the optronic system integrating the matrix.

[0005] STATE OF THE ART

[0006] Array sensors used in infrared (IR) imaging consist of an array of pixels sensitive to infrared radiation, which is arranged at the focal plane of an optical system. Such an array is also known as a focal plane array.

[0007] As is known, the pixels of the matrix subjected to a uniform infrared flux and the same integration time will not have the same response. There are offset disparities (shift of the ordinate at the origin, proportional to the dark noise) and gain disparities from one pixel to another, and these spatial disparities or non-uniformities must be corrected to improve the image quality. In addition, these spatial disparities vary with time.

[0008] We know how to correct this dispersion of individual pixel responses in order to improve the final quality of the image.

[0009] To do this, gain and offset matrices or tables are established for the image sensor in question, which provide, for each pixel, the offset and gain correction to be applied to each pixel value of an image acquired by the pixel matrix.

[0010] In this paper, we are particularly interested in establishing the gain table, which requires being able to image a scene at two different temperatures. Indeed, since the offset depends on the integration time, we cannot use the integration time as a variable to calculate the gain.

[0011] A gain table is obtained during a factory calibration phase by taking two infrared images at two different temperatures. This uses uniform flux emitting sources known as black bodies, electronically controlled to emit at different temperatures.

[0012] One problem is that during the life of an infrared image sensor the spatial non-uniformities can vary over time for several reasons: mechanical (aging of parts), stability of the material sensitive to IR radiation, stability of the cold machine associated with the sensor, etc.

[0013] Therefore, the gain table obtained at the factory no longer allows non-uniformities to be correctly corrected, and the sensor's performance then deteriorates over time.

[0014] To overcome this problem, it is known in optical systems with scanning mirrors to integrate one or more black bodies outside the imager field, with dedicated electronics to obtain uniform emission fluxes at different temperatures. And when the gain table needs to be recalculated, the scanning mechanism is used in one direction to allow the observation of the black body(ies) by the imager. The gain table is then calculated in the same way as in the factory from the image of uniform fluxes at different temperatures. However, these scanning systems tend to disappear because they are too expensive and complex. The integration of the black body(ies) and associated electronics to control their emission temperature is cumbersome and induces power consumption.

[0015] Another solution is to periodically recalibrate the sensor on a dedicated bench. However, this is not always possible and requires disassembling the sensor.

[0016] PRESENTATION OF THE INVENTION

[0017] The invention proposes to overcome at least one of these drawbacks.

[0018] To this end, the invention proposes, according to a first aspect, a method for in situ calibration of a pixel matrix of an infrared sensor, said infrared sensor comprising an optical system, a pixel matrix arranged in the focal plane of the optical system, the optical system making it possible to focus radiation on the pixel matrix, the method comprising the following steps implemented in a processing unit connected to the infrared sensor, the pixel matrix being characterized by a pixel gain table:

[0019] - acquisition by the pixel matrix of a first sequence of images of the radiation of a black screen corresponding to a scene at a first temperature; - acquisition by the pixel matrix of a second sequence of images of the radiation of the partially reflected pixel matrix corresponding to a scene at a second temperature;

[0020] - processing the first and second image sequences so as to update the gain table by calculating a correction representative of the aging and response of the pixels of the pixel matrix.

[0021] The invention, according to the first aspect, is advantageously supplemented by the following characteristics, taken alone or in any of their technically possible combinations:

[0022] - the method comprises a step of configuring the optical system of the infrared sensor so as to acquire the first sequence of images or the second sequence of images;

[0023] - the method comprises a step of rotating the optical system around its optical axis during the acquisition of the second sequence of images;

[0024] - the processing step comprises: determining an intermediate gain table defined by TdG(i,j)=mean(mean(Sensi)). / Sensi(i,j) with Sensl= (moy2-moy1) moy1 and moy2 being the averages of the images of each of the first and second image sequences; and i and j the position indexes of the corresponding pixel in the pixel matrix; determining the current gain table, called TdGc in the following manner TdGc = TdG u _BF + TdG u _HF / TdGi nit _HF * TdG k _HF with: TdG u _BF the low frequency components of the factory gain table, TdG u _HF the high frequency components of the factory gain table; TdGi nit HF a gain table determined when the infrared sensor is first put into service.

[0025] The invention proposes, according to a second aspect, an infrared sensor comprising an optical system, a pixel matrix arranged in the focal plane of the optical system, the optical system making it possible to focus radiation onto the pixel matrix, said system comprising means for switching the optical system into a nominal acquisition mode in which radiation from an observed scene is imaged or into a calibration mode in which the optical system is configured to image the radiation from a black body or the radiation from the pixel matrix partially reflected, said sensor comprising a processing unit configured to implement a method according to the first aspect of the invention. The invention, according to the second aspect, is advantageously supplemented by the following characteristics, taken alone or in any of their technically possible combinations:

[0026] - the optical system comprises a device reflecting incident radiation, the optical system being configured so that in calibration mode said reflecting device partially images the radiation from the pixel matrix;

[0027] - the optical system comprises a four-mirror device, said device being able to switch from the nominal acquisition mode to the calibration mode so that in nominal mode the first mirror comprises a reflecting face parallel to the reflecting face of a second mirror which is perpendicular to the reflecting face of a third mirror which is parallel to the reflecting face of a fourth mirror, in calibration mode, according to a first position of the optical system, the first mirror is arranged to reflect towards the second mirror the radiation of a black body coming from a black-painted plate arranged between the first mirror and the fourth mirror or according to a second position of the optical system to reflect the radiation coming from the reflecting device arranged between the fourth mirror and the first mirror;

[0028] - the reflective device is a cube corner matrix arranged on a plate;

[0029] The invention proposes, according to a third aspect, a computer program product comprising code instructions for implementing a method according to the first aspect of the invention, when the latter is executed by a computer.

[0030] PRESENTATION OF FIGURES

[0031] Other characteristics, aims and advantages of the invention will emerge from the following description, which is purely illustrative and non-limiting, and which must be read in conjunction with the appended drawings in which:

[0032] - figure 1,

[0033] - figure 2 and

[0034] - figure 3 illustrates an infrared sensor according to the invention;

[0035] - Figure 4 illustrates a method of obtaining a factory gain table;

[0036] - Figure 5 illustrates a method for updating a gain table during the life of the product according to the invention. In all the figures, similar elements bear identical references.

[0037] DETAILED DESCRIPTION Sensor structure

[0038] Figure 1 illustrates an infrared sensor 1 comprising an optical system 2 and a pixel matrix 3 arranged in the focal plane 4 of the optical system 2, which is a matrix cooled to a temperature typically between -195°C and -120°C. Such cooling makes it possible to eliminate parasitic currents such as dark or thermal currents. The optical system 2 is configured to image the image of a scene 5 at infinity in a nominal operating mode of the sensor 1. The optical system 2 has an optical axis Z.

[0039] The optical system 2 comprises in particular a mirror device 21 making it possible to bring radiation from a scene observed by an input mirror M1 towards a pupil 22 of the optical system 2 which focuses the radiation towards the focal plane 4. Between the mirror device and the pupil 22 are arranged one or more lenses 23. The mirror device 21 is orientable around the optical axis Z which passes through the centers of the mirrors M1 and M4 for the position illustrated in FIG. 1. In addition, the first mirror M1 can rotate around a local axis Y perpendicular to the optical axis Z and which passes through the centers of the mirrors M1 and M2 for the position illustrated in FIG. 1.

[0040] A processing unit 6 is connected to the pixel matrix 3 and to the optical system 2 and is configured to implement a calibration method which will be described below.

[0041] The pixel matrix 3 is characterized by a gain table. This gain table makes it possible to apply a gain correction to the image acquired by the pixel matrix 3. This gain table is updated during the life of the sensor during a calibration process which will be described later.

[0042] In this respect, the infrared sensor 1 can operate in two modes: a nominal mode in which it can image a scene at infinity. This is the normal operation of the sensor; a calibration mode in which the gain table of the pixel matrix is ​​updated. In this respect, the sensor 1 comprises a unit 7 for switching from the nominal mode to the calibration mode. In particular, this unit 7 can configure the optical system 2 for each of the modes.

[0043] The optical system 2 comprises a mirror device 21 which makes it possible to bring the radiation towards the pupil 22. The device 21 more particularly comprises four mirrors M1 to M4, where the first mirror M1 is an input mirror in the optical system 2, which captures the radiation of a scene which is to be imaged by the pixel matrix. As indicated previously, all four mirrors M1, M2, M3, M4 are rotatable around the optical axis Z; furthermore, according to the invention, at least the first mirror M1 can also be pivoted around the Y axis.

[0044] It will be specified that the Y axis is a local axis, defined in the rotating frame of the mirror device 21. More precisely, M1 rotates around the local Y axis which is defined by the straight line connecting the centers of the mirrors M1 and M2 and this local Y axis rotates when the 4 mirrors of the mirror device 21 rotate around the Z axis.

[0045] It is understood that depending on the operating mode, the mirrors M1, M2, M3, M4 can be arranged differently.

[0046] In the nominal mode illustrated in Figure 1, the optical system 2 is configured to allow imaging of a scene at infinity, facing the sensor. The configuration of the device 21 of four mirrors corresponding to this operating mode is as follows: the first mirror M1 is parallel to a second mirror M2, a third mirror M3 is perpendicular to the second mirror M2, a fourth mirror M4 is parallel to the third mirror M3 and is perpendicular to the first mirror M1. The reflecting faces of these four mirrors are such that the radiation of a scene 5 at infinity enters through the input mirror M1 then goes towards M2 then goes towards M3 then goes towards M4 to go into the pupil 22 passing through the lenses 23.

[0047] The optical system 2 further comprises two additional elements: a black body CN, typically a device such as a plate, which is painted black; and a reflecting device MCC. These two elements are positioned in the space between the mirror M4 and the mirror M1, so as to each be accessible in a respective position of the first mirror M1. That is to say that the mirror M1 can be pivoted to one side or the other relative to its nominal position (that of figure 1). The reflecting device MCC and the black body CN are involved in the calibration mode. They are arranged in the mirror device 21, in a fixed manner, and so as to be outside the acquisition field in the nominal operating mode, and fulfill their role, in conjunction with the mirror M1, in calibration mode.

[0048] In this calibration mode, the optical system 2 successively takes two different positions which are used for consecutive steps of the calibration method, each position corresponding to a configuration in which the sensor can image an infrared source, a first position in which the sensor can acquire the image of an infrared source at a first temperature constituted by the black body CN which is at the internal temperature of the sensor, judiciously placed between the mirrors M1 and M4, the input mirror M1 being suitably oriented to receive and transmit the infrared radiation emitted by the black body;and a second position where the image of an infrared source at the temperature of the pixel matrix is ​​virtually formed by means of the reflecting device MCC judiciously placed between the mirrors M1 and M4, the mirror M1 being suitably oriented to receive and transmit the infrared radiation reflected by the reflecting device.;

[0049] As schematically illustrated in top view in Figure 2 (schematic top view), in the calibration mode, the black body CN and the reflecting device MCC are each arranged in the free space between the mirrors M1 and M4 such that for the first position, the first input mirror M1 in the optical system 2 which captures the radiation of a scene to be imaged by the pixel matrix is ​​pivoted on a first side, relative to its position in nominal mode, around the local axis Y (Figure 1) so that its reflecting face looks at the black body CN. According to this position, the first mirror M1 is pivoted on one side around the local axis Y to be oriented so that its line of sight A1 looks at the black body CN.For the second position also illustrated in Figure 2 the first mirror M1 is pivoted to the other side around the local axis Y, relative to its position in nominal mode, so that its reflecting face looks at the reflecting device MCC, its line of sight A2 being oriented accordingly.

[0050] Thus, the optical system 2 makes it possible to image in calibration mode two scenes at two different temperatures simply by acting on the position M1 to image the black body CN which is at the internal temperature of the sensor, or to image an image of the focal plane itself, by means of the reflective device MCC, the focal plane usually being at a different temperature, controlled by a device for cooling the pixel matrix. Advantageously, the reflective device MCC is a matrix of cube corners arranged on a plate. A cube corner is a well-known element, used in many applications such as: bicycle wheel retro-reflectors; roadside retro-reflectors; telemetry.

[0051] The fundamental property of a cube corner is to reflect back in the direction of incidence the rays reflected by the three surfaces of each unit element. In order to increase the reflected energy by making the reflector compact and robust, several cube corners are placed side by side, forming a matrix of cube corners. The reflected rays are parallel to the incident rays, but not superimposed. The three reflections on the faces of each cube corner shift the beam. In the context of usual applications, often at great distances, the backscattered and captured energy in return is sufficient for the desired function.

[0052] Thus, in the context of the sensor illustrated in the figures, in the calibration mode, when the first mirror M1 is in the first position, it reflects the radiation of the black body, positioned behind the fourth mirror M4, this makes it possible to obtain an emitting source at a first controlled temperature corresponding to the internal temperature of the product.

[0053] Furthermore, when the first mirror M1 is in the second position, it reflects the radiation from the cube corner matrix MCC to the focal plane matrix. This is in fact mainly the radiation from the focal plane matrix reflected by the cube corner matrix MCC. Indeed, the shift of the reflected beams implies that not all the rays emitted by the pixel matrix re-enter the pixel matrix: not all the rays coming from the pixel matrix are reflected back to the pixel matrix.

[0054] Since the pixel matrix is ​​cooled to a temperature very different from the internal temperature of the product, the image resulting from the reflection by the MCC cube corner matrix corresponds to the image of a source at a second controlled temperature, which is the temperature of the pixel matrix, different from the internal temperature. In a realistic practical example, the internal temperature of the product is between -40°C and +110°C while the temperature of the pixel matrix is ​​between -195°C and -120°C. It is thus possible to acquire two sequences of images corresponding to two emitting sources at two different temperatures, which are the internal temperature of the optical system and the specific temperature of the imager which is cooled.

[0055] This is achieved at a lower cost in terms of physical integration into the optronic system both in terms of size and electronics, because the black body CN and the reflective element MCC which make it possible to produce two infrared sources emitting at different temperatures, are placed in the free space between the mirrors M1 and M4, so as to be out of the acquisition field in operational mode of operation of the optical head, when the mirror M1 is in its nominal position, parallel to the mirror M4. These additional elements CN and MCC are passive elements, not requiring associated electronics. The electronic position control mechanism of the mirror M1 allowing in the calibration mode to image these two infrared sources does not pose any difficulties in terms of production or integration.Typically, a position control device (orientation) of the mirror M1 with respect to the local Y axis is initialized with a nominal position value, a position value corresponding to the mode of capture of the black body radiation (first acquisition sequence in the calibration phase) and a position value corresponding to the mode of capture of the radiation of the reflecting device MCC (second acquisition sequence in the calibration phase). The position of the mirror M1 is thus controlled according to the current operating mode.

[0056] Such an optical system allows, as we will describe, to carry out an in situ calibration of the sensor, allowing during the life of the sensor, without the need for disassembly, to update the pixel gain table used nominally to improve the image quality, taking into account the effects of pixel aging, but also, as we will explain later, the specific variations of so-called atypical pixels.

[0057] Before describing in more detail the calibration method according to the invention, it should be remembered that the sensor is calibrated on a bench in the factory, in particular to establish a factory gain table, TdG u .

[0058] This factory gain table TdGu is obtained in the following usual manner, schematically illustrated in Figure 4. The sensor is positioned so as to image a black body located in front of the sensor. Note that the acquisition is carried out during an integration time Tint. This integration time is preserved during the calibration mode according to the invention and in nominal mode. In this way, the variability of the offset with the integration time is overcome.

[0059] A first sequence of images of the black body at a first temperature TCN1 is acquired (step E01), for example at 15°C. A second sequence of images of the black body brought to a second temperature TCN2 is acquired (step E2), for example at 35°C.

[0060] The images obtained are matrix images which provide for each pixel of the pixel matrix, the quantity of signal integrated by the pixel during the integration time Tint. In the following, the term image is used to designate the operation - of image acquisition - of the matrix sensor, while the term matrix used alone, or the expression image matrix, designates the set of signals acquired by the pixels during an image acquisition which therefore form a representation of the image. The dimension of the matrices representing the images corresponds to the dimension of the pixel matrix of the sensor.

[0061] Each image sequence is then averaged in order to remove the Poisson temporal noise in particular (steps E03 and E04). We thus calculate respectively for the first and second image sequences, an average matrix, respectively moylo ​​and moy2o, of identical sizes, corresponding to the dimension of the acquired images, for respectively the first temperature TCN1 and the second temperature TCN2.

[0062] The TdG factory gain table u is then calculated from these two average matrices moylo ​​and moy2o. This is step E05 of figure 4, and this calculation is written as follows:

[0063] (EQ0): TdGu(i,j)=meani(meanj(Senslo)) / Senslo(i,j) with Senslo(i,j)= (moy2o(i,j)-moy1o(i,j)), corresponding to a normalization of the sensitivities (slopes) of the pixels, with moylo ​​and moy2o, averages on the time axis.

[0064] Each pixel i,j has a gain correction which is equal to the ratio between the difference of the spatial averages of the images moylo, moy2o and the difference of the values ​​of the pixel i,j in these two average images, moy1o(i,j), moy2o(i,j).

[0065] This being explained, the calibration method according to the invention makes it possible to update, that is to say to correct, the factory gain table, to take into account variations in pixel responses during the life of the product, the infrared sensor, whether these are the effects of pixel aging, or even atypical variations in pixels of the sensor's pixel matrix.

[0066] In terms of pixel gain, these variations translate in practice into a dispersion of the high-frequency response of the pixels.

[0067] The method according to the invention makes it possible to correct the high-frequency components of the factory gain table over the life of the sensor, to take into account the evolution of the high-frequency response of the pixels over time.

[0068] As schematically illustrated in Figure 5, when the sensor is put into service, the processing unit contains the factory gain table noted TdG u (step E10) which will have been obtained as described above (figure 4). We separate the low frequency and high frequency components of this TdG table u (steps E12, E13, figure 5), for example using a polynomial basis of the Zernike or Legendre type. We obtain a factory gain table "BF" and a factory gain table "HF" which we denote TdG u _BF and TdG u _HF.

[0069] The method according to the invention then makes it possible, in a step E14, to calculate a current gain table TdG c, from the factory gain table "BF" and a factory gain table "HF" corrected for pixel variations since the product was put into service. We note TdG u _HF_corr this high frequency corrected gain table. The current gain table TdG c is the one that is applied to improve the quality of the images provided by the sensor.

[0070] The method according to the invention for calculating the high-frequency corrected gain table and updating the current gain table of the sensor is more particularly illustrated in Figure 6. It is carried out in situ, using the calibration mode of the optical system allowing two sources of different temperatures to be imaged in a simple manner.

[0071] It comprises a sequence comprising steps of acquiring image sequences of the two different temperature sources, by ad hoc control, i.e. by pivoting, around the local Y axis, the mirror M1 as explained previously, and thanks to the integrated passive elements CN and MCC in the mirror device 21, and steps of processing these image sequences, to calculate a gain table on the basis of a formula EQ1, similar to the formula EQO described above in relation to FIG. 4, but applied to the image sequences of the sources of different temperatures produced in situ, by the sensor. This sequence of steps including the calculation of a gain table, which is called an intermediate gain table, because it qualifies the response of the sensor at a moment in the operational life of the sensor, is carried out a first time, at commissioning, corresponding to the start of the operational life of the optronic device integrating the sensor.We obtain an initial intermediate gain table, which we filter to keep only the high-frequency components of this table. We note this table TdGinitJHF, which serves as a reference for measuring the gain variation to come during the life of the sensor.

[0072] Then, when we need to update the factory gain table, we execute a new iteration of the sequence of steps to calculate a new intermediate gain table, to which we apply the same filtering functions used for the initial intermediate gain table to keep only the high-frequency components. We denote this table TdG k _HF. The subscript k is used to signify that it is calculated during a k-th iteration of the calculation step, with k >1 .

[0073] The HF factory gain table can then be corrected for pixel gain variations since commissioning, based on these two tables TdGinit_HF and TdG k _HF.

[0074] Thus, in the invention, the current gain table TdG c applied in operation by the sensor corresponds to the sum of the BF part of the factory gain table, noted TdG u _BF, with the HF part corrected from the factory gain table, noted TdG u _HF_corr (Figure 5). It is precisely this last term which is calculated and updated, in an integrated manner, in situ, during the life of the sensor, by the calibration mode of the invention.

[0075] In more detail and with reference to Figure 6, the method thus comprises a first sequence of a step of acquiring two image sequences, corresponding to two infrared sources at two different temperatures, and a step of calculating a so-called intermediate gain table noted TdGi. This first sequence is common for the initialization phase E8M and the following iterative calibration phases E8 k .

[0076] For the acquisition step, the optical system is for example first positioned so that the black body is imaged (step E1). A first sequence of images of the black body CN is acquired (step E2). This corresponds to the image of an emitting source at a first temperature, which is the internal temperature of the sensor.

[0077] The optical system is then positioned so that the MCC cube corner matrix is ​​imaged (step E3). A second image sequence that corresponds to the radiation from the pixel matrix partially reflected by the MCC matrix is ​​then acquired (step E3).

[0078] Advantageously, during this step the optical device 21 is rotating around its Z axis (step E3'). By this rotary movement, the HF disturbances of the edges of the cubic corners of the MCC matrix are compensated by averaging them.

[0079] Then, the images of each sequence are averaged in order to remove the Poisson temporal noise in particular (steps E5 and E6). We thus obtain for the first and second sequence of images, respectively a matrix avg1 and a matrix avg2 which are of sizes identical to the dimension of the images.

[0080] We then calculate in step E7, the intermediate gain table, noted TdGi ,. This intermediate gain table is obtained in a similar way to the factory gain table, using the formula EQO but applied to the average matrices avg1 and avg2 calculated on the first and second image sequences from steps E1 to E4 of the method of the invention. This formula is denoted EQ1, which if we use the usual Maltab™ notations is then written:

[0081] (EQ1): TdGi(i,j)=mean(mean(Sensi)). / Sensi(i,j) with Sensi = (moy2-moy1)

[0082] The resulting intermediate gain table is then filtered to retain only the high frequency components.

[0083] During the first iteration, when the product is put into service, of this sequence of steps E1 to E7 of acquisition and processing of the first and second image sequences, EQ1 calculation and filtering, we thus obtain the high-frequency components TdGini t_HF of the intermediate gain table, representative of the sensor response to commissioning (step E8M). This first iteration allows a calibration of the correction (update) function of the method according to the invention, similar to a measurement of the transfer function of the in situ calibration system.

[0084] Then, at each subsequent iteration of the sequence of steps E1 to E7, triggered during the life of the sensor, we obtain at step 8 k the high-frequency TdG components k JHF of the HF intermediate gain table representative of the current response of the sensor at that time.

[0085] The gain variation (HF component) since commissioning can thus be calculated by reference to the initial HF intermediate gain table TdGi nit _HF.

[0086] The current gain table can then be updated, in a step E9, which only takes into account the variation in pixel gain, since the sensor was put into service, as follows:

[0087] (EQ2): TdG c = TdG c = TdG u _BF + TdG u _HF_corr with TdG u _HF_corr = TdG u _BF + TdG u _HF / TdG init _HF * TdG k _HF where

[0088] - TdG u _BF the low frequency components of the factory gain table,

[0089] - TdG u _HF the high frequency components of the factory gain table.

[0090] - TdGi nit _HF and TdG k _HF the high frequency components of the initial intermediate gain table and at iteration k, calculated in situ in the product.

[0091] This current gain table TdGc is the one used to correct the images acquired in nominal mode by the sensor, until the next update.

[0092] The method of the invention is simple to use and easy to integrate. In addition, the implemented gain correction only takes into account the HF components of the gain tables, this advantageously makes it possible to overcome various low-frequency disturbances induced by the elements used for calibration.

[0093] Indeed, the use of the MCC cube corner matrix to obtain the second sequence of images corresponding to the radiation of an infrared source at a temperature which is the temperature to which the pixel matrix is ​​cooled, induces low-frequency disturbances in the images. These are

[0094] - shadowing effects, induced by the MCC matrix given its position in the optical field,

[0095] - disturbances linked to the double passage of radiation in the optical device 2.

[0096] These undesirable BF disturbances linked to the use of the cube corner matrix thus remain without effect on the calibration according to the invention.

[0097] The use of the cube corner matrix MCC also causes high-frequency disturbances, through the reflection effects on the edges of the cube corners of the MCC matrix. It is advantageously proposed to cancel these HF disturbances in the image signals, by imposing a rotary movement on the optical system 2 including all 4 mirrors, the body CN and the cube corner matrix MCC, around the optical axis Z during the acquisition of the second sequence of images, so as to average these HF disturbances (see figure 1). This corresponds to step E3' of the method of the invention described previously and illustrated in figure 6.

[0098] Finally, the method according to the invention, which makes it possible to correct the gain of the pixels in situ during the life of the sensor, naturally makes it possible to deal with the case of atypical pixels, the gain of which is likely to vary during the life of the product, independently of the effects of aging. At each iteration of the method, the gain table is updated, thus dealing with this case of possible atypical pixels.

[0099] There is no longer any need to search for possible atypical pixels at the factory testing stage, which involves complex measurements with probability calculations to detect atypical pixels and declare them defective or not, with a criterion generally more severe than necessary.

[0100] Thus, by means of the simple and isovolume integration of two elements, black body and MCC type reflective matrix, in the optical head, it is possible, by simple processing requiring few calculations on the sequences of images obtained, to have a current gain table of the pixels of the sensor which follows the evolution of the pixels during the life of the product, and this in operation, without the need to disassemble the sensor. The correction of the images obtained is thus optimized, taking into account both the variations specific to possible atypical pixels and the effects of aging, and this throughout the life of the products.

Claims

DEMANDS 1. Infrared sensor (1) comprising an optical system (2), a pixel array (3) disposed in the focal plane (4) of the optical system (2), the optical system (2) for focusing radiation onto the pixel array (3), said system comprising means (5) for switching the optical system (2) into a nominal acquisition mode in which radiation from an observed scene is imaged or into a calibration mode in which the optical system (2) is configured to image blackbody radiation or partially reflected radiation from the pixel array, the optical system comprising a four-mirror device and a reflecting device (MCC) for incident radiation,the device being such that in nominal mode the first mirror (M1) comprises a reflective face parallel to the reflective face of a second mirror which is perpendicular to the reflective face of a third mirror (M3) which is parallel to the reflective face of a fourth mirror (M4), in calibration mode, according to a first position of the optical system (2), the first mirror (M1) is arranged to reflect towards the second mirror (M2) the radiation from a black body emitted from a black-painted plate disposed between the first mirror (M1) and the fourth mirror (M4) or according to a second position of the optical system (2) to reflect the radiation emitted from the reflecting device (MCC) disposed between the fourth mirror (M4) and the first mirror (M1), the reflecting device (MCC) partially imaging the radiation from the pixel matrix said sensor comprising a processing unit (6) configured to implement, in calibration mode,the following steps: - acquisition (E2) by the pixel matrix of a first sequence of images of the radiation from a black screen (CN) corresponding to a scene at a first temperature; - acquisition (E4) by the pixel matrix of a second sequence of images of the radiation from the partially reflected pixel matrix corresponding to a scene at a second temperature; - processing (E5, E6, E7, E8, E9) of the first and second image sequences in order to update the gain table by calculating a correction representative of the aging and response of the pixels of the pixel matrix.

2. Sensor according to claim 1, wherein the processing unit is configured to control a rotation of the optical system around its optical axis (Z) during the acquisition of the second image sequence.

3. Sensor according to any one of the preceding claims, wherein the reflective device is a cube-cornered matrix arranged on a plate.

4. A sensor according to any one of the preceding claims, wherein the processing unit is configured to implement the processing step comprising: - the determination (E7) of an intermediate gain table defined by TdG(i,j)=mean(mean(Sensi)). / Sensi(i,j) with Sensl= (moy2-moy1) moy1 and moy2 being the means of the images of each of the first and second image sequences; and i and j the position indices of the corresponding pixel in the pixel matrix - the determination (E9) of the current gain table, called TdGc, as follows TdGc = TdG u _BF + TdG u _HF / TdG init _HF * TdG k _HF With - TdG u _BF the low-frequency components of the factory gain table, - TdGu _HF the high-frequency components of the factory gain table; - TdGini t HF a gain table determined at the first commissioning of the infrared sensor.

5. Product computer program comprising code instructions to implement a method according to any one of claims 1 to 4, when executed by a computer.

Images