Method for correcting residual optical defects

Abstract

According to one aspect, the invention relates to a method (400) for correcting residual optical defects of an image acquired by a camera with a particular focal length, the acquired image being corrected for defects of a sensor of the camera, wherein the residual optical defects relate to the particular focal length and are different from the defects of the sensor, the method being carried out on the basis of a first calibration table corresponding to an amplitude of the residual optical defects at the particular focal length.

Description

DESCRIPTION TITLE: Method for correcting residual optical defects TECHNICAL FIELD OF THE INVENTION

[0001] The technical field of the invention is that of correcting optical defects in an image acquired by an IR camera having a cooled detector.

[0002] The present invention relates to a method for correcting residual optical defects in an image acquired by a cooled and previously calibrated IR camera. TECHNOLOGICAL BACKGROUND OF THE INVENTION

[0003] In an infrared camera, each optical element generates a spatially homogeneous parasitic thermal flux. In theory, the parasitic flux emitted by the camera's cryogenic detector can be imaged very diffusely by the detector or not imaged at all.

[0004] Thus, the parasitic thermal flux of the camera can be modeled as a spatially homogeneous background noise directly linked to the ambient temperature of the camera's optical elements, excluding the detector. This phenomenon is due to a radiative transfer between the internal optical elements of the camera up to its detector. [Fig. 1] thus shows a graph of the temperature T of the parasitic flux received by the detector on the central line of the detector (and therefore on the central line of the image it produces), the abscissas corresponding to the pixels Pi_c of the central line. The graph is constant due to the homogeneity of the background noise and is equal to an observed temperature Te.

[0005] However, in some cases, the optical configuration of the infrared camera causes reflections of the thermal flux emitted by the camera's cryogenic detector, which receives and images its own cold thermal flux. Since a thermal flux is greater the higher the temperature of a thermal source to be imaged by the camera, the cold flux from the cryogenic detector acts as a jump in the thermal flux received by the camera, the flux from the detector being colder than the flux from the thermal camera.

[0006] As a result, the thermal transfer emitted by a source to be imaged is altered and the flow received by the detector is composed of two flows: the thermal flow from the camera and the cold flux from the detector. The cold flux from the detector thus replaces the homogeneous parasitic hot flux from inside the camera.

[0007] This results in a temperature gradient on the detector, resulting in a "gray level" gradient and therefore a non-uniformity of parasitic noise on the image. The shape of the non-uniformity depends on the shape of the detector and the optical configuration of the camera. The intensity of the non-uniformity depends on the different temperatures in the camera, for example the temperature of the detector and the ambient temperature, as well as the integration time for example. [Fig. 2] is a graph representing the temperature of the flux received by the detector, at the pixels Pi_c of the central line of the detector. In particular, [Fig. 2] is a graph representing the temperature T of the flux received by the central line of the detector, part of which has a temperature corresponding to the temperature Te of the flux received by the camera, and part of which has a parasitic temperature Tp depending on the temperature Td of the flux emitted by the detector and the temperature Te.In particular, the temperature Tp can be equal to: Tp = (1 - Q) * Te + Q * Td, with Q corresponding to the proportion of stray flux from the camera replaced by stray flux from the detector. The higher the detector reflection, the higher Q.

[0008] When using a cooled infrared camera, it is possible to correct the non-uniform detector noise mentioned above by a one-point calibration process for example.

[0009] The one-point calibration of a cooled infrared camera consists of adjusting a gain table G and an offset table O, the adjustment being carried out from a defocused image or from a diaphragm (from the English "shutter") used to obtain a supposedly uniform image on the sensor.

[0010] In particular, the response of a pixel to a flux can be modeled by a linear function characterized by a slope coefficient that represents a gain and an ordinate at the origin that represents an offset (commonly called offset). Each pixel has its own gain and its own offset. Consequently, when receiving a uniform light flux, the gray levels generated by the pixels differ, which results in the appearance of non-uniformities on the image generated by the camera.

[0011] [Fig. 3] shows an example of the response of a pixel P1 and the response of a pixel P2 distinct from P1, the response of a pixel corresponding to gray level values ​​N as a function of the luminance L which depends on the temperature T of the received flux. Thus, in [Fig. 3], each pixel response has a different steering coefficient - and therefore a different gain, and a different ordinate at the origin - and therefore a different offset.

[0012] In order to eliminate pixel non-uniformities, it is necessary that the pixels have the same gain and offset. A correction is made by applying the one-point calibration method, from a gain table G and an offset table O, each of which has the same dimensions as the detector matrix.

[0013] The gain table G is expressed as a table of coefficients, each coefficient being multiplied to a pixel of the same position. The gain table G is determined in the factory before commissioning the camera, for example. After its application, all pixels have the same response to a given flux variation and their characteristic functions therefore have the same direction coefficient. The gain table can be determined by performing the first-order partial derivative of two images taken at two different scene temperatures. In particular, the partial derivative is performed with respect to the luminance parameter, which depends on the temperature.

[0014] The offset table O is additive relative to the flow received by each of the pixels, and is therefore expressed in the form of a table of factors applied to each of the pixels, determined during operation of the camera.

[0015] A shift table O is obtained from the image of a scene acquired in front of a diaphragm and noted diaphragm, multiplied by the gain table G. In particular, the shift table O is obtained from the following formula:

[0016] O — Idîaphragm * G — Average aperture * G). Each pixel of Average (I diaphra g me * G) includes the average of all pixels in the ^diaphragm * G matrix. After its application, all pixels will have the same response to a given flux in the configuration where the calibration was carried out.

[0017] [Fig. 4] is a block diagram of a one-point calibration method 100, based on a gain table G determined according to the method described above. The method 100 includes a step 101 of shuttering the camera. In particular, the optical defects seen and corrected by the camera are not the same depending on the shuttering method chosen, the shuttering being able to be carried out by defocusing, or by placing the camera diaphragm in front of the entire optical path or in front of the sensor.

[0018] The method further comprises a step 102 of calculating and applying an integration time. The method then comprises a step 103 of acquiring an image of the diaphragm Idiaphra me then a step 104 of constructing the offset table O determined according to the previous formula. Steps 101 to 104 correspond to the camera calibration phase.

[0019] Furthermore, the method 100 comprises a step 105 of acquiring any image Im by an operator with the integration time, then a step 106 of multiplying the acquired image by the gain table G and finally a step 107 of subtracting the offset table O, which makes it possible to obtain in theory a final image Imf free from intrinsic defects of the detector and parasitic fluxes due to the camera.

[0020] In particular, if step 101 of the calibration method is performed with a diaphragm in front of a specific optical field, all defects of the "optical reflection" type will be perfectly corrected. However, if there is a change in the camera field, the O offset table is no longer valid, i.e. it no longer corrects the "optical reflection" type defects because the optical defects vary with the field. In the case of known cameras, the calibration by the O offset table is performed in a specific optical configuration called the calibration field during which the flux coming from the scene is averaged on the sensor. Thus, the O offset table corrects the defects seen only in this specific optical configuration.Non-uniformity or offset defects (differences between the ordinates at the origin of the pixel responses) in the usual fields are not corrected because it is currently not known how to calculate an offset table in real time on the usual fields, at the risk of embedding the visible scene in the offset table. In particular, if a defect appears locally in a field where the offset table O has not been made, it is not corrected. This phenomenon is all the more visible if the camera has a continuous zoom and can image several fields.

[0021] If step 101 of the calibration method is performed by "defocusing", that is to say if the optics are moved to defocus the image at the extreme (calibration field), optical reflection defects - if any - are no longer present. Thus, when the O offset table is produced in this case, the "optical reflection" type defects are not taken into account and only the intrinsic defects of the detector are corrected. When the camera "refocuses" the image, the image is calibrated but the reflection defects are still visible because they were not present, i.e. not visible, when the O offset table was generated.

[0022] As a result, non-uniformities or residual optical defects, caused by parasitic flows internal to the camera, may remain present on the image despite a one-point calibration as described previously.

[0023] There is therefore a need to correct residual optical defects in an acquired and calibrated image. SUMMARY OF THE INVENTION

[0024] The invention provides a solution to the problems mentioned above, by making it possible to determine the residual optical defects of an image acquired at a given and calibrated focal length. The residual optical defects of the image are determined from at least one calibration table comprising amplitude values ​​of the residual defect for the given focal length and an adaptation factor making it possible to take into account the differences in acquisition conditions between the calibration table and the acquired image to correct the residual optical defects.

[0025] One aspect of the invention relates to a computer-implemented method of correcting residual optical defects in an image acquired by a camera with a given focal length, the acquired image being calibrated, the residual optical defects being relative to the given focal length, the method comprising: Obtaining a correction of the residual optical defects of the acquired image, the correction depending on: a first calibration table corresponding to an amplitude of the residual optical defects at the given focal length and; a first adaptation factor between the first calibration table and the acquired image, Obtain a second image corrected for residual optical defects from the acquired calibrated image and the correction, by subtracting the correction from the acquired image.

[0026] The term "acquired image being calibrated" means an image on which a calibration method such as a one-point calibration is applied. Thus, on a calibrated image, the intrinsic defects of the camera detector are corrected. In particular, if the offset table of the calibration method was obtained from a diaphragm in front of a specific optical field, all defects of the "optical reflection" or "parasitic defects" type are also corrected in the calibration field only and not in other fields. If the offset table is obtained by defocusing, the defects of the "optical reflection" or "parasitic defects" type are not corrected in any field. Thus, optical reflection defects or parasitic optical defects are called residual optical defects in the following.

[0027] In particular, defects of the "optical reflection" or "parasitic defects" type are Narcissus defects, due to non-uniformities created from parasitic thermal fluxes emitted by the camera detector, which reflect on at least one optic of the camera and are imaged on the detector. Residual optical defects depend on the field (or equivalently the focal length) of the camera. Residual optical defects are distinct from intrinsic defects of the camera detector and are modeled as spatial additive noise. Thus, residual optical defects are present in the acquired image after its calibration.

[0028] In particular, a given field of view of the camera corresponds to a given focal length F'd of the camera lens. The given field of view is defined by an angle called field angle ad expressed according to the following formula: ad = 2*arctan(^-), with d the length of an edge or a diagonal of an image acquired by the camera.

[0029] By means of the invention, the residual optical defects of an acquired and calibrated image are corrected from the first calibration table associated with the same field (or equivalently with the same focal length) as that of the image acquired and calibrated in the factory for example. In particular, the adaptation factor makes it possible to adjust the amplitude of the first calibration table in order to bring the optical defects closer together. residual optical defects of the acquired image to the residual optical defects calibrated in the factory. The first adaptation factor can take into account the integration time of the camera during the acquisition of the image to be corrected and that used in the factory for the calibration table, the amplitude of the residual optical defects depending on the integration time for example. Thus, the invention allows an increase in the quality of restitution of the information of the acquired image by homogenizing the acquired image, unlike the state of the art in which only the defects due to the pixels of the detector are corrected. The reading of the acquired image by a user is improved and details are more visible when it is necessary to observe a distant target for example. Advantageously, the invention makes it possible to eliminate any optical defect modeled as spatial additive noise in the field considered, regardless of its shape or geometry, without altering the acquired image.

[0030] In addition to the characteristics which have just been mentioned in the preceding paragraph, the method according to one aspect of the invention may have one or more additional characteristics among the following, considered individually or according to all technically possible combinations.

[0031] According to one embodiment, the correction comprises a product of the first adaptation factor and the first calibration table.

[0032] According to one embodiment, the residual optical defects are parasitic thermal flux defects associated with the given focal length and are modeled by spatial additive noise. Advantageously, any optical defect relating to a given focal length and modelable by spatial additive noise can be corrected by the invention.

[0033] According to one embodiment: the first calibration table is associated with a calibration integration time of the camera and with an internal calibration temperature of the camera and, the image is acquired for a given integration time of the camera and for a given internal temperature of the camera and the first adaptation factor is determined from the calibration integration time, the given integration time, the internal calibration temperature and the given internal temperature of the camera, the first adaptation factor being determined from the calibration integration time, the given integration time, the calibration internal temperature and the given internal temperature. Advantageously, the first adaptation factor makes it possible to take into account the different acquisition conditions of the image to be corrected and of the calibration table. In particular, the first adaptation factor takes into account the integration time of the camera on which the amplitude of the residual optical defects depends. Thus, the first adaptation factor varies when the integration time varies.

[0034] According to one embodiment, the first adaptation factor is defined as a product: From a ratio of the given integration time and the calibration integration time, and From a ratio of the given internal temperature and the calibration internal temperature.

[0035] According to one embodiment, the first adaptation factor is defined as a product: From a ratio of the given integration time and the calibration integration time, and From a ratio of a luminance obtained from the given internal temperature and a luminance obtained from the calibration internal temperature.

[0036] According to one embodiment, the correction further depends on: a second calibration table corresponding to a variation in the amplitude of the residual optical defects as a function of a variation in the internal temperature of the camera for the given focal length, and a second adaptation factor between the second calibration table and the acquired image.

[0037] According to one embodiment, the first calibration table is associated with a calibration integration time of the camera and an internal calibration temperature of the camera, and the image is acquired for a given integration time of the camera and a given internal temperature of the camera and according to which: the first adaptation factor is determined from the calibration integration time and the given integration time, the second adaptation factor is determined from the first adaptation factor, the given internal temperature and the calibration internal temperature.

[0038] According to one embodiment, the first calibration table is associated with a calibration integration time of the camera and an internal calibration temperature of the camera, and the image is acquired for a given integration time of the camera and a given internal temperature of the camera and: the first adaptation factor is determined from the calibration integration time and the given integration time, the second adaptation factor is determined from the first factor, from a luminance obtained from the given internal temperature and from a luminance obtained from the internal calibration temperature.

[0039] According to one embodiment, the first adaptation factor and the second adaptation factor are respectively equal to a first initial value and a second initial value and obtaining the correction comprises: Obtaining an intermediate image by subtracting an intermediate correction from the acquired image, the intermediate correction being determined from the first adaptation factor equal to the first initial value, the first calibration table, the second adaptation factor equal to the second initial value and the second calibration table, Optimization of a criterion relating to the intermediate image, by obtaining a first optimal adaptation factor and a second optimal adaptation factor, the correction of residual optical defects being determined from the first optimal adaptation factor and the second optimal adaptation factor.

[0040] According to one embodiment, the residual optical defects are modeled as a radial geometry defect and the optimization of the criterion comprises a step of applying a filter adapted to the radial geometry of the defect to the intermediate image to obtain a filtered intermediate image and the criterion corresponds to a variance of the filtered intermediate image. For example, the filter adapted to the radial geometry is a gradient filter or a Laplacian filter.

[0041] Another aspect of the invention relates to a computer program product comprising instructions which, when the program is executed by a computer, cause the latter to implement the method according to the invention.

[0042] Another aspect of the invention relates to a computer-readable medium comprising instructions which, when executed by a computer, cause the computer to implement the method according to the invention.

[0043] The invention and its various applications will be better understood by reading the following description and examining the accompanying figures. BRIEF DESCRIPTION OF THE FIGURES

[0044] The figures are presented for information purposes only and in no way limit the invention.

[0045] [Fig. 1] is a uniform graph of the temperature T of the parasitic flux received by the detector on the pixels Pi_c of the central line of the detector in the prior art.

[0046] [Fig. 2] is a non-uniform graph of the temperature T of the parasitic flux received by the detector on the pixels Pi_c of the central line of the detector in the prior art.

[0047] [Fig. 3] represents the responses of two pixels to a luminance from a thermal flux, corresponding to graphs of the gray levels at each pixel as a function of the luminance received by the pixel in the prior art.

[0048] [Fig. 4] is a block diagram representing a state-of-the-art method for correcting non-uniform defects on an image, corresponding in particular to detector defects.

[0049] [Fig. 5] is a block diagram showing the steps of a method for determining the first calibration table.

[0050] [Fig. 6] is a block diagram showing the steps of a method for determining the second calibration table.

[0051] [Fig. 7] is a block diagram showing the steps of a correction method according to the invention.

[0052] [Fig. 8] is a block diagram representing the sub-steps of a step of determining a correction of residual optical defects of the correction method according to the invention.

[0053] [Fig. 9] shows three graphs corresponding to an amplitude of optical defects on the center line of an image.

[0054] [Fig. 10] is a diagram showing a device configured to implement the method according to the invention. DETAILED DESCRIPTION

[0055] The invention relates to a method for correcting residual optical defects in an image acquired by an infrared camera, preferably the infrared camera having a detector cooled to a constant temperature.

[0056] The camera is located, for example, in an aircraft or in an observation viewfinder located in a land vehicle and acquires, for example, images which can be displayed in real time or later to a user of the aircraft or land vehicle, for example.

[0057] An "infrared camera" is a camera configured to measure and record heat waves and infrared radiation emitted by a body or object, in order to generate so-called infrared image sequences. The camera is, for example, a Short-Wave Infrared (SWIR), Long-Wave Infrared (LWIR) or Midwave Infrared (MWIR) camera.

[0058] In particular, the camera includes a zoom lens, that is, a lens with a variable focal length that can be adjusted by a user. In the following, we will confuse "focal length" and "focal length". As described above, the camera's field of view is inversely proportional to the focal length.

[0059] In particular, the infrared camera comprises an infrared detector, formed for example from a matrix of FPA (Focal Plane Array) type detectors.

[0060] In particular, the images acquired by the camera can be grayscale images.

[0061] When the camera acquires an image of a scene, it includes parasitic optical defects or reflection defects resulting from the detector, the defects depending on the camera's field of view. In particular, the camera lens includes a front lens on which parasitic optical thermal fluxes coming from the camera's cooled detector are reflected and imaged on said detector. When the lens configuration is modified to change the lens focal length and therefore the camera's field of view, the optical path of the parasitic fluxes is modified, which modifies their impact on the image acquired by the camera. In other words, the optical defects resulting from parasitic thermal fluxes depend on the camera's focal length and therefore vary when the focal length varies. These optical defects are called parasitic optical defects or internal reflection optical defects and can be of any geometry, for example a radial or square geometry.In particular, the geometry of parasitic optical defects has rotational symmetry.

[0062] For example, when the camera is calibrated using method 100, the steps of which are shown in [Fig. 4] and have been described previously, the optical defects of internal reflection (due to the detector) are corrected for a given field if the shutter is made by a diaphragm, or are not corrected (because they are not visible) if the shutter is made by defocusing. Thus, as described previously, optical defects of internal reflection depend on the field of the camera, the one-point calibration method described in method 100 does not allow them to be eliminated if the shift table (or offset) was obtained for a different field or was obtained by defocusing (and not by the use of a physical diaphragm).

[0063] The offset table used in the one-point calibration method and the real-time acquired image are each obtained for a camera integration time and camera internal temperature that may be equal to or different from the camera internal temperatures and integration times, respectively, when acquiring a real-time image.

[0064] In particular, the amplitudes of the detector's internal reflection optical defects vary depending on the internal temperature and the camera integration time.

[0065] The term "internal camera temperature" refers to the temperature measured near the optics of the camera lens, for example, near the front lens of the lens. In the following, any internal camera temperature value is, for example, within the range [-20; +70] °C.

[0066] In the remainder of the writing, the optical reflection defects present in an image after calibration, in particular after the application of a one-point calibration method, are called residual optical defects. The correction method according to the invention thus aims to correct the residual optical defects, for each given focal length F'd and each temperature and integration time, without having to repeat a one-point calibration method (method 100). According to the invention, the correction of these defects is carried out from a first calibration table 01 d and a second calibration table O2d, each calibration table being associated with the given focal length F'd.

[0067] The first calibration table 01 d associated with the given focal length F'd has the same size as the size of a matrix of the camera detector, and therefore the same size as the image to be corrected, and corresponds to a matrix of amplitudes of the residual optical defects for the given focal length F'd. In particular, each coefficient of the first calibration table corresponds to an amplitude of the residual optical defects for a given position of a pixel of the camera detector.

[0068] The second calibration table O2d has the same size as the size of the camera detector matrix, and therefore the same size as the image to be corrected, and corresponds to a matrix modeling variations in the amplitudes of the residual optics as a function of variations in the internal temperatures of the camera for the given focal length F'd. In particular, each coefficient of the second calibration table corresponds to a value of variation in the amplitude of the residual optical defects for a given position of a pixel of the camera detector.

[0069] Examples of determining the first and second calibration tables are described in the rest of the article.

[0070] In particular, for each focal length F'i, the first calibration table 01 i associated with the focal length F'i can be determined according to a method for determining the first calibration table. [Fig. 5] presents a block diagram of the steps of the method 200 for determining the first calibration table 01 i associated with the focal length F'i. In particular, each calibration table 01 i is associated with the internal calibration temperature of the camera and the calibration integration time of the camera.

[0071] The method 200 comprises a step 201 of acquisition by the camera of an image Isj of a thermal flux emitted by a homogeneous and isotropic source, for the focal length F'i of the camera, the internal calibration temperature of the camera and the calibration integration time of the camera which are stored by a user in a memory external to the camera for example.

[0072] A "homogeneous and isotropic source" is understood to mean a source emitting, for example, a thermal flux (or thermal radiation) similar to the thermal flux emitted by a black body, i.e. a homogeneous and isotropic thermal flux, whose radiation depends only on the temperature and not on the directions in space, for example. Advantageously, the use of a source emitting isotropic radiation makes it possible to have a homogeneous scene and to observe only the optical defects of the camera.

[0073] In practice, the source can be a heating plate covered entirely with black paint, and having a uniform, constant temperature that can be adjusted by a user. Thus, the source can be likened to a black body.

[0074] The heat flux emitted by the source covers the entire pupil of the camera lens.

[0075] The method 200 comprises a step 202 of calibrating the image Isj according to the method 100 of one-point calibration of the camera in which the shuttering step corresponds to a defocusing. This makes it possible to correct the intrinsic defects of the detector but does not correct the reflection defects of the latter. The one-point calibration is carried out from the gain table G and the offset table O described previously. After the calibration step, an image of the calibrated source Is2j is obtained from the image of the source Isj, such that: I S2 i = G * l s i - O

[0076] The image Is2j obtained corresponds to the image of the source Isj corrected for the detector defects and includes residual optical defects as mentioned previously, i.e. residual optical defects corresponding to parasitic thermal fluxes of the camera lens relative to the focal length Fï

[0077] The method 200 further comprises a step 203 of obtaining a third image Issj resulting from the subtraction of an average image of the image Is2j, denoted Average(Is2j), from the calibrated image Is2j. In particular, Average(Is2j) corresponds to an image of the same size as Is2j and each coefficient of which corresponds to the average of all the pixel values ​​of Is2j.

[0078] So, I S3 _i = I S 2_i - Average (I S2 i ).

[0079] According to a first embodiment, the first calibration table 01 i associated with the focal length F'i corresponds to Issj and can therefore be noted: 01; = I S3

[0080] Thus, the first calibration table 01 i can be obtained at the end of step 203.

[0081] According to a second embodiment, the method 200 comprises an optional step 204 of spatial filtering of the third image Issj to obtain a filtered image F(ls3j).

[0082] The filtering may for example correspond to a spatial filtering of the image Issj in the case where intrinsic defects of the detector remain even after the calibration step 201, for example because of one or more defective pixels of the detector. In this second embodiment, the first calibration table 01 i may be equal to F(ls3j).

[0083] According to a third embodiment, the first calibration table 01 i is obtained at the end of optional steps not shown in the method 200, described below.

[0084] The method 200 may comprise a first optional step of carrying out steps 201, 202 and 203 at several different times t and obtaining a plurality of images (Is3jt)t>o

[0085] The method 200 may comprise a second optional step of averaging the images Is3i_t to obtain an average image Mean(ls3jt)t>o. In particular, the averaging step 206 corresponds to an averaging of each value of pixel of each image Issjt having the same position in the image. In other words, each coefficient at a given position of the image Mean(ls3jt) t>o corresponds to an average of the values ​​of the pixels of the same position of each image Issjt.

[0086] The method 200 may include a third optional step of obtaining the first calibration table 01 i by subtracting the image Mean(ls3jt) from the image Issjt. Thus, 01; = I S3i - Average(I S3 it ).

[0087] In particular, the third embodiment advantageously makes it possible to remove temporal noise present in the image Is3j, the temporal noise degrading the quality of the image by giving it a granular appearance.

[0088] According to an embodiment not shown, the optional steps are applied to the image F(IS3_i) following step 204.

[0089] The method 200 for obtaining the first calibration table 01 i associated with the focal length F'i can be carried out for a plurality of M focal lengths which makes it possible to obtain M first calibration tables (OIO^^M each associated with a different focal length F'i, the index i being a natural integer. In particular, each first calibration table among the M first calibration tables is obtained for a different focal length but with the same internal calibration temperature of the camera and the same calibration integration time of the camera.

[0090] In particular, each first calibration table may be compressed before being saved in a memory for use. The compression method must be adapted to the spatial morphology of the residual optical defects represented by each first calibration table, in order to avoid any degradation of the quality of the first calibration table.

[0091] Concerning the second calibration table O2i, for each focal length F'i, the second calibration table O2i associated with the focal length F'i can be determined according to a method for determining the second calibration table. [Fig. 6] presents a block diagram of the steps of the method 300 for determining the second calibration table O2i associated with the focal length F'i. In particular, each calibration table O2i is associated with the camera calibration integration time.

[0092] The method 300 comprises a step 301 of acquisition by the camera of a first image 11 sj of a thermal flux emitted by the source defined in the method of determining the first calibration table. In particular, the acquisition step 301 is carried out for the focal length F'i of the camera, at an internal calibration temperature T1 calibration of the camera and with the calibration integration time T1 calibration of the camera which are stored by a user in a memory external to the camera for example or in a memory of the camera.

[0093] The method 300 further comprises a step 302 identical to step 202, the step 302 being applied to the image 11 sj and making it possible to obtain the calibrated image 11 S2j. The calibrated image I1 s2j corresponds to the image 11 sj corrected for the detector defects and comprising residual optical defects as described previously. The residual optical defects correspond to parasitic thermal fluxes of the camera lens for the focal length F'i.

[0094] The method 300 further comprises a step 303 of obtaining an image I1 s3j resulting from the subtraction of an average image of the image I1 s2j, denoted Average(H s2j) from the calibrated image I1 s2j. In particular, Average(I1s2j) corresponds to an image of the same size as I1 s2j and each coefficient of which corresponds to the average of all the pixel values ​​of 11 S2j. Thus, 11 ssj = I1 S2 i - Average(It S2 .

[0095] The method 300 further comprises a step 304 of varying the internal temperature of the camera, to obtain an internal temperature T2calibration of the camera different from T1 calibration. The variation of the internal temperature of the camera can be achieved by placing the camera in a temperature-adjustable oven, for example.

[0096] The method 300 comprises a step 305 of acquisition by the camera of a second image I2sj of the thermal flux source defined in the method 200, for the focal length F'i of the camera, the internal calibration temperature T2 e camera calibration and camera calibration integration time.

[0097] The method 300 further comprises a step 306 identical to step 202, the step 306 being applied to the image I2sj and making it possible to obtain the calibrated image I2s2j. The calibrated image I2s2j corresponds to the image I2sj corrected for the detector defects and comprising residual optical defects as mentioned previously.

[0098] The method 300 further comprises a step 307 of obtaining an image I's3j resulting from the subtraction of an average image of the image I2s2j, denoted Average(I2s2j) from the calibrated image I2s2j. In particular, Average(I2s2j) corresponds to an image of the same size as I2s2j and each coefficient of which corresponds to the average of all the pixel values ​​of I2s2j. Thus, I2 S3 = I2 S2 - Average (I2 S2 ).

[0099] The method 300 comprises a step 308 of obtaining the second calibration table O2i from the image I 1 s3j, the calibration temperature T1 calibration, the image I2s3j and the calibration temperature T2 calibration.

[0100] In particular, the second O2i calibration table makes it possible to estimate the variation of the residual optical defects as a function of the variation in internal temperature. An interpolation is carried out between the variation of the residual optical defects and the variation in internal temperature, the interpolation order being chosen according to predetermined criteria. Indeed, the higher the interpolation order, the more the quality of the images to be corrected for residual optical defects is improved because the relationship between the variation of the residual optical defects and the variation in internal temperature is more precise. However, the higher the interpolation order, the more complex the method 300 is to implement because a plurality of blackbody image acquisitions are required.

[0101] In particular, the variation of residual optical defects can depend linearly and indirectly on the variation of luminance due to the variation of temperature. Indeed, the variation of residual optical defects depends linearly on the variation of luminance which depends on the temperature.

[0102] Thus, the second O2i calibration table is obtained according to the following formula: 02; = — - S3 -' f - 1 - - with L(T1 calibration) the integrated luminance generated by the temperature T1 calibration and L(T2calibration) the integrated luminance generated by the temperature T2calibration.

[0103] Thus, the integrated luminance is related to the internal temperature T that the camera integrates (captures). The gray level captured by the camera depends on a multitude of sources. For the purposes of interpolation, a simulation of the variation of gray levels captured by the detector as a function of the variation in internal temperature is carried out. The determination of the luminance as a function of the temperature (denoted L(T1) and L(T2)) is for example obtained via an infrared camera simulation method.

[0104] In particular, the integrated luminance takes into account a fixed temperature T of a scene integrated by the camera and passing through the optics, the internal luminance of the camera (i.e., the photons leaving the camera and reaching the detector), and the dark current of the camera detector.

[0105] When the infrared camera is placed in front of a scene of fixed temperature T, the internal temperature of the camera is varied in order to observe the evolution of the gray level variation as a function of the variation of the internal luminance of the camera.

[0106] The integrated luminances are obtained from a resolution of the Planck equations in a context applied to the camera, that is to say by taking into account the variation of gray level as a function of the variation of the internal luminance of the camera.

[0107] Using luminance values ​​as a function of temperature is advantageous because the luminance of a pixel in an image is modeled linearly as a function of its gray level, for any temperature. Thus, since luminance is linear with gray levels, the simulation is valid for all temperatures.

[0108] The method 300 for obtaining the second calibration table O2i associated with the focal length F'i can be carried out for a plurality of M focal lengths (F'i)i <m> ce which allows obtaining M second calibration tables each associated with a different focal length F'i. In particular, each first calibration table among the first M calibration tables is obtained for a different focal length but with the same camera calibration integration time.

[0109] Thus, the method 200 represented in [Fig. 5] and the method 300 represented in [Fig. 6] respectively make it possible to obtain first and second calibration tables used in the method 400 for correcting residual optical defects according to the invention, the steps of which are represented in [Fig. 7],

[0110] [Fig. 7] in fact presents a block diagram of the method 400 for correcting residual optical defects according to the invention. In particular, the method 400 aims to correct the residual optical defects of an Irrid image acquired for a given focal length F'd, by the camera described previously, and corrected for the detector defects using the one-point calibration method for example.

[0111] The Irrid image to be corrected for residual optical defects is acquired by the camera for an integration time td and an internal temperature Td, stored in a camera memory or a memory external to the camera for example.

[0112] In particular, in order to correct these residual optical defects, the first calibration table 01 d associated with the given focal length F'd is used in method 400.

[0113] In addition, the second calibration table O2d associated with the given focal length F'd is used in method 400.

[0114] The correction method 400 comprises a step 401 of obtaining a correction Corrd of the residual optical defects of the image Irrid, from the first calibration table 01 d and the second calibration table O2d. The correction Corrd is an image of the same size as the image to be corrected, determined from the first calibration table 01 d, a first adaptation factor K1 d relating to the first calibration table 01 d, the second calibration table O2d and a second adaptation factor K2d relating to the second calibration table O2d.

[0115] In particular, the first adaptation factor K1d makes it possible to adjust the amplitude of the first calibration table 01 d to the image to be corrected, the first calibration table 01 d and the Irrid image to be corrected each comprising the amplitude of the residual optical defects for the focal length F'd but for different acquisition conditions. The first adaptation factor K1d is a real strictly greater than 0.

[0116] In particular, the second adaptation factor K2d makes it possible to adjust the amplitude of the second calibration table O2d to the image to be corrected, in order to take into account the effects of internal temperature variations of the camera on the amplitude of residual optical defects. The second adaptation factor K2d is a real strictly greater than 0.

[0117] For example, the correction Corrd is equal to K1 d* 01 d + K2d* O2d, in the case of an interpolation of order 1.

[0118] In particular, the first adaptation factor K1d and the second adaptation factor K2d can be obtained according to several embodiments described below.

[0119] In a first embodiment for obtaining the Corrd correction, the variation in the amplitude of the residual optical defects of the internal temperature of the camera is modeled as uniform over the entire image to be corrected and all the coefficients of the second calibration table O2d are thus equal to 1. In this embodiment, the second adaptation factor K2d is zero.

[0120] Thus, Corrd = K1d* 01 d . In this first embodiment, the optical defects to be corrected in the image have a deterministic formula to calculate their amplitude which corresponds to the reflection of the detector on the optics (commonly called the Narcissus effect). The deterministic formulas are based on the physical nature of the residual optical defects.

[0121] In this first embodiment, the first calibration table 01 da was obtained for the integration time Tetration and the internal temperature of the camera Tetration. The first adaptation factor K1d depends on Tetration, Tetration, the integration time td of the camera during the acquisition of the image to be corrected, the internal temperature Td of the camera during the acquisition of the image to be corrected.

[0122] In particular, we consider the following quantities: L(Tétaionnage), the luminance integrated by the camera for an object having a temperature Tétaionnage and L(Td) the luminance integrated by the camera for an object having a temperature Td. K1d can be equal

[0123] In particular, as previously specified, for any temperature T, the luminance L as a function of the temperature T is obtained according to Planck's formula.

[0124] In particular, in this first embodiment, K1d may be equal to a function or a polynomial of the parameter - — — — for example, the polynomial being 1 calibration preferably of order m greater than or equal to 2.

[0125] In a second embodiment of obtaining the Corrd correction, the variation in the amplitude of the residual optical defects of the internal temperature of the camera is modeled as non-uniform over the entire image to be corrected. Thus, the first adaptation factor K1 d and the second adaptation factor K2d are unaffected, and the second calibration table O2d includes coefficients not necessarily equal to 1. In this second embodiment, the optical defects to be corrected in the image have a deterministic formula for calculating their amplitude which corresponds to the reflection of the detector on the optics (corresponding to the Narcissus effect). The deterministic formulas are based on the physical nature of the residual optical defects.

[0126] In this second embodiment, K1 d is for example equal to — - — ^calibration and K2d is for example equal

[0127] In a third embodiment, step 401 of obtaining the correction Corrd comprises sub-steps 4011 and 4012 shown in [Fig. 8]. In this embodiment, the factors K1d and K2d have respectively any initial value K1d_o and K2d_o, for example equal to 0. This embodiment makes it possible to correct optical defects which do not admit formulas making it possible to calculate their amplitudes, for example the vignetting defect typically. This embodiment can also be used for the Narcissus effect.

[0128] Sub-step 4011 is a step of obtaining an intermediate image lm(interm)d' = Irrid - Corrd_ interm, with Corrd_ interm an intermediate correction equal to (K1 d * 01 d + K2d * O2d), with the factors K1d and K2d equal to their respective initial values ​​K1d_o and K2d_o.

[0129] Sub-step 4012 is a step of optimizing the first factor K1d and the second factor K2d in order to reduce a criterion on the image lm(interm)d'. In particular, sub-step 4012 makes it possible to find optimal values ​​respectively for K1d and K2d, also noted K1d_o P tet K2d_o P t such that the criterion on the image lm(interm)d' is respected.

[0130] For example, if the residual optical defects are modeled as low-frequency additive spatial noise, the lm(interm)d' image is filtered to eliminate high frequencies and the criterion to be reduced or even minimized is a variance of the filtered lm(interm)d' image, which makes it possible to minimize the effects of the residual optical defects.

[0131] In particular, the image lm(interm)d is spatially filtered so as to eliminate any defects that may be present other than optical reflection defects. It is necessary that the filtering does not alter said optical reflection defects.

[0132] Optical reflection defects can be of a spatial high-frequency (HF) nature, of a spatial low-frequency (LF) nature or, for example, a mixture of the two natures. Therefore, in the classic cases encountered, a classic high-frequency or low-frequency filtering via a convolution matrix, for example, is not optimal.

[0133] In particular, optical reflection defects are defects with rotational symmetry. If optical reflection defects have a radial geometry with rotational symmetry, an example of a spatial filter is a radial filter that will retain only spatial frequencies with rotational symmetry.

[0134] In particular, optimization is carried out for example using the gradient descent algorithm.

[0135] Thus, at the end of step 4012, optimal values ​​are obtained for the first factor K1 d and for the second factor K2d, such that the condition on the criterion of the image lm(interm)d' is respected.

[0136] Referring again to [Fig. 7], the method 400 comprises a step 402 of correcting residual optical defects in the acquired Imd image.

[0137] Correction step 402 is performed by subtracting the Corrd correction from the Imd image, to obtain an Imd' image corrected for residual optical defects. Thus, Imd' is equal to Imd - Corrd.

[0138] Thus, the user observes the Imd' image of better quality than the Imd image and free of residual optical defects.

[0139] [Fig. 9] represents three graphs of amplitudes of defects present in a central line of an image. The y-axis thus represents the amplitude of defects in an image and the x-axis represents the pixels Pi_c of the central line of the image.

[0140] In particular, a first graph represents the amplitude of defects in a central line of an IM image acquired by the camera, without any correction. This includes detector faults and optical faults due to stray flux from the camera optics.

[0141] A second graph represents the magnitude of defects in an IM image, obtained from a one-point calibration method applied to the IM image. Thus, the magnitude of defects has decreased compared to the first graph, as defects due to the detector have been eliminated. However, residual optical defects as described previously remain.

[0142] A third graph represents the amplitude of the defects of an IM image obtained after applying the residual optical defect correction method according to the invention. Thus, the amplitude of the optical defects is uniform and approaches a zero value unlike the other graphs. Thus, the correction method according to the invention makes it possible to reduce or even eliminate the residual optical defects of the acquired images.

[0143] Another aspect of the invention relates to a device configured to implement the method of correcting residual optical defects of an image according to the invention.

[0144] [Fig. 10] represents the device according to the invention, according to one or more embodiments of the invention.

[0145] In these embodiments, the device comprises a computer 500, comprising a memory 501 for storing instructions which, when implemented by a processor of the computer, cause the processor and therefore the computer to implement the correction method.

[0146] The computer 500 further comprises a circuit 502. This circuit may be, for example, a processor capable of interpreting instructions in the form of a computer program, an electronic card whose steps of the method of the invention are described in the silicon, or even a programmable electronic chip such as an FPGA chip (for “Field-Programmable Gate Array” in English).

[0147] The computer 500 comprises an input interface 503 for receiving the image acquired by the camera, for example, and corrected for detector defects, and an output interface 504 for providing the image corrected for residual optical defects, for example. Finally, the computer may comprise, to allow easy interaction with a user, a screen 505 and a keyboard 506. Of course, the keyboard is optional, especially in the case of a computer in the form of a touch pad, for example.

[0148] Of course, the present invention is not limited to the embodiments described above as examples; it extends to other variants.< / m> ​

Claims

CLAIMS

1. A computer-implemented method (400) for correcting residual optical defects in an image (Irrid) acquired by a camera with a given focal length (F'd), the acquired image (Irrid) being calibrated, the residual optical defects being relative to the given focal length (F'd), the method (400) comprising: - Obtain (401) a correction (Corrd) of the residual optical defects of the acquired image (Irrid), the correction (Corrd) comprising a product: o of a first calibration table (01 d) corresponding to an amplitude of the residual optical defects at the given focal length (F'd) and; o of a first adaptation factor (K1d) between the first calibration table (01 d) and the acquired image (Irrid), - Obtain (402) a second corrected image (Im'd) of the residual optical defects from the acquired calibrated image (Irrid) and the correction (Corrd), by subtracting the correction (Corrd) from the acquired image (Irrid).

2. Method (400) according to the preceding claim according to which the residual optical defects are parasitic thermal flux defects associated with the given focal length and are modeled by spatial additive noise.

3. A method (400) according to any preceding claim wherein: - the first calibration table (01 d) is associated with a calibration integration time (t calibration) of the camera and an internal calibration temperature (T calibration) of the camera and, - the image (Irrid) is acquired for a given integration time (td) of the camera and for a given internal temperature (Td) of the camera, and according to which the first adaptation factor (K1d) is determined from the calibration integration time (t calibration), the given integration time (td), the internal calibration temperature (Tetaionnage) and the given internal temperature (Td) of the camera. Tl

4. Method (400) according to the preceding claim according to which the first adaptation factor (K1 d) is defined as a product: - From a ratio of the given integration time (td) and the calibration integration time (t calibration), and - A ratio of the given internal temperature (Td) of the camera and the internal temperature (Testing) of calibration.

5. Method (400) according to claim 3 according to which the first adaptation factor (K1d) is defined as a product: - From a ratio of the given integration time (td) and the calibration integration time (t calibration), and - From a ratio of a luminance (L(Td)) obtained from the given internal temperature (Td) of the camera and a luminance (L(Tetration)) obtained from the internal temperature (Tetration) of calibration.

6. A method according to any one of claims 1 or 2 wherein the correction (Corrd) further depends on: - a second calibration table (O2d) corresponding to a variation in the amplitude of the residual optical defects as a function of a variation in the internal temperature of the camera for the given focal length F'd and; - a second adaptation factor (K2d) between the second calibration table (O2d) and the acquired image (Irrid).

7. Method (400) according to claim 6 according to which the first calibration table (01 d) is associated with a calibration integration time (t calibration) of the camera and with an internal calibration temperature (T calibration) of the camera, and the image (Irrid) is acquired for a given integration time (td) of the camera and a given internal temperature (Td) of the camera and according to which: the first adaptation factor (K1d) is determined from the calibration integration time and the given integration time, the second adaptation factor (K2d) is determined from the first adaptation factor (K1d), the given internal temperature (Td) and the calibration internal temperature (Td).

8. Method (400) according to claim 6 according to which the first calibration table (01 d) is associated with a calibration integration time (t calibration) of the camera and with an internal calibration temperature (T calibration) of the camera, and the image (Irrid) is acquired for a given integration time (td) of the camera and a given internal temperature (Td) of the camera and according to which: - the first adaptation factor (K1d) is determined from the calibration integration time and the given integration time, - the second adaptation factor (K2d) is determined from the first adaptation factor (K1 d), from a luminance (L(Td)) obtained from the given internal temperature (Td) and from a luminance (L(Tetration)) obtained from the calibration internal temperature (Tetration).

9. Method (400) according to claim 6 according to which the first adaptation factor (K1 d) and the second adaptation factor (K2d) are respectively equal to a first initial value (K1d_o) and a second initial value (K2d_o) and according to which the obtaining (401) of the correction comprises: - Obtain (4011 ) an intermediate image lm(interm)d' by subtracting an intermediate correction (Corrdjnterm) from the acquired image, the intermediate correction (Corrdjnterm) being determined from the first adaptation factor (K1d) equal to the first initial value (K1d_o), the first calibration table (01 d), the second adaptation factor (K2d) equal to the second initial value (K2d_o) and the second calibration table (O2d), - Optimization (4012) of a criterion relating to the intermediate image lm(interm)d', by obtaining a first optimal adaptation factor (K1 d_o Pt) and a second optimal adaptation factor (K2d_o P t), the correction of residual optical defects being determined from the first optimal adaptation factor (K1d_o P t) and the second optimal adaptation factor (K1 d_opt).

10. Method (400) according to the preceding claim according to which the residual optical defects are modeled as a radial geometry defect and the optimization of the criterion comprises a step of applying a filter adapted to the radial geometry of the defect to the intermediate image to obtain a filtered intermediate image and the criterion corresponds to a variance of the filtered intermediate image.

11. A computer program product comprising instructions which, when the program is executed by a computer, cause the computer to implement the method according to one of claims 1 to 10.

12. A computer-readable medium comprising instructions which, when executed by a computer, cause the computer to implement the method according to one of claims 1 to 10.

Images