Method for using an infrared camera

Abstract

A method for using an infrared camera, comprising the following steps: - determining (214) a first reference integration time, - obtaining a first offset table associated with the first reference integration time, - determining (218) a second reference integration time different from the first reference integration time, - obtaining a second offset table associated with the second reference integration time, - constructing (222) at least one interpolation table representing the variations of the offset as a function of the integration time from at least the first offset table and the second offset table, - acquiring (230) an image during a current integration time, and - correcting non-uniformities in the image using the current integration time and the interpolation table. Figure to be published with the abbreviation: 2

Description

Title of the invention: Method of using an infrared camera Technical field of the invention

[0001] The invention relates to a method for calibrating an infrared camera, in particular of the cooled infrared camera type. State of the prior art

[0002] An infrared camera typically comprises a housing comprising a lens and optics allowing the entry of infrared radiation and the guidance of said radiation towards a detector comprising a plurality of photosensitive cells, or wells. The camera further comprises a computer control system receiving the signal measured by the detector and processing the images.

[0003] Such a camera allows the acquisition of two-dimensional images comprising a grid of pixels each having a signal level, or gray level to be displayed, determined by the quantity of infrared radiation captured by the wells associated with the pixels of the camera detector, during a predetermined exposure time, or integration time.

[0004] Such a camera is for example used to continuously observe a scene by taking images of it periodically, with an exposure time chosen so that the hot spots of the scene are correctly observable without saturating the detector. Generally, this time is calculated to obtain an average signal level of the pixels of the image equal to 50% of the maximum perceptible level, which corresponds to an average filling of 50% of the wells of the pixels of the detector. This level is calculated on the acquired image, then applied to the following image. This operating mode was chosen as a compromise between an acceptable value of thermal sensitivity and the preservation of a high dynamic range on the image, so as not to immediately saturate any hot spots which are often points of interest.

[0005] The calibration of a cooled infrared camera is done by adjusting a gain G and an offset O, which can in particular be implemented from a defocused image or a diaphragm, both applied to the pixel levels of the images taken.

[0006] At first order, the response of the pixel to a luminous flux can be characterized by an affine function characterized by a direction coefficient also called gain and an ordinate at the origin also called offset. Each pixel has its own gain and its own offset. Consequently, in front of a uniform luminous flux, the gray level generated by the pixels will be different from one pixel to another, resulting in the appearance of non-uniformities on the image. To eliminate the non-uniformities, it is necessary that the pixels have the same gain and the same offset. To do this, we apply a gain correction table G and an offset correction table O, tables which therefore have the same dimensions as the detector matrix.

[0007] The gain table G is multiplicative relative to the signal level of each of the pixels, and is therefore expressed in the form of a table of factors to be applied (one factor / pixel) to each of the pixels which is determined in the factory before the camera is put into service. After its application, all the pixels can have the same response to a given flux variation: the same steering coefficient.

[0008] The offset table O is additive relative to the levels of 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. After its application, all the pixels will have the same response to a given flux. The offset table corrects, for example, defects linked to the dark current, which is a specific signal generated by each pixel of the detector even in the absence of light illumination, the response of which in gray levels depends on the integration time.

[0009] The offset O is generally determined when the camera is in operation. Such a correction is particularly necessary each time the integration time is modified to obtain a satisfactory offset correction in the image, since it corresponds to the integration time of this image.

[0010] Such a modification of the integration time is particularly desirable during changes in the scene studied which cause hot spots to appear or disappear or modify them sufficiently so that the contrast scale of the image must be changed.

[0011] [Fig. 1] illustrates a method of using such a cooled infrared camera, known from the state of the art. This method comprises offset compensation, implemented following a change in integration time or simply when commissioning the camera.

[0012] The method of use begins with a step 100 of starting up the infrared camera, followed by a calibration step 110 comprising compensation of the offset O.

[0013] The calibration step 110 comprises sub-steps of defocusing 112, determining the integration time 114, offset compensation 116 and refocusing 118.

[0014] The defocusing 112 and the refocusing 118 correspond respectively to the installation and removal of a diaphragm in front of the lens of the infrared camera, said diaphragm being suitable for blocking the arrival of infrared radiation on the detector from the outside and the observed scene. The objective is to decorrelate the flux received by the detector from the observed scene.

[0015] During the step of determining the integration time 114, a new integration time is calculated from previous images of the scene or from pre-recorded data, said new integration time aiming to obtain an average filling of 50% of the wells of the detector as explained above.

[0016] During the offset compensation step 116, an offset image is acquired during the integration time set up in the previous step, and with the diaphragm blocking the entry of infrared radiation into the camera. The measured signal is therefore only made up of the offset due in particular to the dark current. An offset table is then directly formed from the offset image, and intended to be subtracted from the subsequent images to compensate for this defect.

[0017] The method of using the infrared camera then comprises steps of acquiring an image 120, correcting the gain G 122 using the gain table determined in the factory, correcting non-uniformities 124 using the offset table O previously determined for the integration time used, and obtaining a final image 126.

[0018] The method finally comprises a test step 130 during which an evolution of the observed scene is evaluated from the newly acquired image, in order to determine whether it is necessary to modify the integration time. This test step 130 may in particular be based on the calculation of an average filling rate of the wells of the detector in the image.

[0019] If such a modification is necessary, a new calibration step 110 must be initiated in order to determine a new offset table corresponding to this new integration time.

[0020] A problem associated with this method is that the implementation of the calibration step results in the stopping of the taking of images of the observed scene for a duration generally of the order of 2 to 3 seconds. The camera is unusable during this duration, and this phenomenon occurs each time the integration time is modified.

[0021] It is therefore difficult to regularly adapt the integration time, and the quality of the images suffers when the observed scene varies greatly. The implementation of dynamic control of the integration time is not acceptable in view of the operational conditions with this type of process. Presentation of the invention

[0022] The invention aims to remedy these drawbacks by providing a method of calibrating an infrared camera allowing frequent variation of the integration time without excessively disrupting the operation of the camera.

[0023] To this end, the invention relates to a method of using an infrared camera, comprising the following steps:

[0024] - determination of a first reference integration time,

[0025] - obtaining a first offset table associated with the first integration time of reference,

[0026] - determination of a second reference integration time different from the first reference integration time,

[0027] - obtaining a second offset table associated with the second integration time of reference,

[0028] - construction of at least one interpolation table I representative of the variations of the offset as a function of the integration time from at least the first offset table and the second offset table,

[0029] - acquisition of an image during a current integration time, and

[0030] - correction of non-uniformities of the image acquired from the integration time current and interpolation table.

[0031] Such a method makes it possible, for each pixel of the image, to determine by interpolation the offset adapted for the current integration time. The method thus makes it possible to correct the images acquired by the camera without having to recalibrate it each time the current integration time changes.

[0032] Each interpolation table I may comprise values ​​of a coefficient of a polynomial function describing the variations of the offset as a function of the integration time for each pixel of the detector.

[0033] Such a characteristic allows a simple and precise interpolation of the offset as a function of the integration time.

[0034] The polynomial function may be a linear function, and the interpolation table I may be unique and include leading coefficients.

[0035] Such a characteristic makes it possible to consider a linear variation of the offset with the integration time, which is adapted to the dark current phenomenon and allows simple and rapid interpolation.

[0036] According to one embodiment, the polynomial function may be a function of order greater than 1 and the method may comprise the construction of several interpolation tables.

[0037] The step of correcting the non-uniformities of the image may comprise a first step of correcting the image by subtraction of the first offset table and a step of adjusting the corrected image by subtraction, in particular an additional subtraction, of a product of the interpolation table I by a time difference between the first reference integration time and the current integration time of said image.

[0038] The first offset table, the second offset table and each interpolation table I may each comprise a distinct value for each pixel of the images acquired by the camera.

[0039] In particular, the tables can have the same dimension as the sensor in order to correspond to each pixel, a polynomial coefficient value per table.

[0040] Such a characteristic makes it possible to independently correct each of the pixels in the image, to take into account local variations in the defects causing non-uniformities in the image.

[0041] The method may comprise, after the step of correcting the non-uniformities of the image, a step of obtaining a final image followed by a step of changing the current integration time, the changing step being directly followed by a new step of acquiring a new image with the new integration time.

[0042] Such a feature allows dynamic control of the integration time to adapt it to changes in the observed scene, without immobilizing the camera for recalibration.

[0043] During the changing step, the current integration time can be modified to obtain a desired average filling rate of the pixels of the image.

[0044] Such a characteristic makes it possible to increase the filling of the wells when observing scenes without strong hot spots, making it possible to increase the sensitivity, while reducing the sensitivity when such hot spots are present to avoid saturating the pixels and losing discrimination.

[0045] The invention also relates to a computer program comprising instructions for implementing the aforementioned method, when said instructions are executed by a processor of a processing circuit.

[0046] The invention also relates to a computer-readable recording medium comprising instructions which, when executed by a computer, cause the latter to implement the steps of the method as mentioned above.

[0047] The invention also relates to a device comprising means for implementing the method as mentioned above. In particular, the device may comprise a camera equipped or connected to a processor configured to implement the method as mentioned above. Brief description of the figures

[0048] [Fig.l] [Fig.l] is a schematic representation of a method of using an infrared camera according to the state of the art, and

[0049] [Fig.2] [Fig.2] is a schematic representation of a method of using an infrared camera according to the invention. Detailed description of the invention

[0050] A method of using an infrared camera, in particular an in- cooled infrared, is shown schematically in [Fig.2].

[0051] As described above, the method of use comprises a step 200 of starting up the infrared camera, followed by a calibration step 210 comprising compensation of the offset O.

[0052] The calibration step 210 comprises a defocusing step 212, a first step 214 of determining a first reference integration time, a first step 216 of calculating the offset with this first integration time, a second step 218 of determining a second reference integration time, a second step 220 of calculating the offset with this second integration time, a step 222 of constructing an interpolation table I and a refocusing step 224.

[0053] The defocusing steps 212 and refocusing steps 224 are identical to those previously described.

[0054] The first and second integration times are determined, based on the average level of the image, during the first determination step 214 and the second determination step 218 so as to obtain two filling rates that are different from each other, and significantly separated from each other. For example, the two filling rates obtained with the first integration time and the second reference integration time are separated by at least 25% of the filling capacity of the wells, and advantageously by at least 50%.

[0055] For example, the first integration time is determined in order to have a filling rate of 20% and the second integration time is determined in order to have a filling rate of 80% of the wells.

[0056] In particular, the integration time is calculated from the measurement of the average filling of the pixel wells and the current integration time. Thus, it is possible to obtain the desired average level by proportionality as a function of the integration time.

[0057] During the first calculation step 216, a first reference image is measured with the first reference integration time with a shuttered objective. A first offset table obtained with this first reference integration time is determined and recorded.

[0058] Similarly, during the second calculation step 220, a second reference image is measured with the second reference integration time and a second offset table is determined and recorded.

[0059] During the construction step 222, an interpolation table I is generated from the first offset table and the second offset table and the first and second integration times.

[0060] The interpolation table I makes it possible to express, for each pixel of an image, the variation of the offset level to be applied to the pixel depending on the integration time with which the image was acquired.

[0061] For example, the interpolation table I comprises, for each pixel, a direction coefficient of a linear regression obtained with the offset of the first offset table and the offset of the second offset table for this pixel.

[0062] This direction coefficient is equal to the difference between the offset of the second table and that of the first table divided by the difference between the second reference integration time and the first reference integration time.

[0063] The method of using the infrared camera then comprises steps of acquiring an image 230 during a current integration time, determined for the current scene to be imaged, and which can therefore vary from one image to the next. It is a priori distinct from the first and second reference integration rates.

[0064] The method then comprises a gain correction step 232 implementing the gain table determined in the factory, as previously detailed above, in the part describing the state of the prior art.

[0065] The method then comprises a first correction step 234, during which the first offset table, obtained for the first reference integration time, is subtracted point by point from the image.

[0066] The method then comprises a deviation calculation step 236, during which an integration time deviation is calculated by the difference between the current integration time and the first reference integration time.

[0067] The method then comprises an adjustment step 238, during which, for each pixel of the image, the product of the integration time difference by the direction coefficient associated with said pixel provided by the interpolation table I is subtracted from the level of said pixel.

[0068] Then, the method comprises a step 240 of obtaining a corrected final image.

[0069] The method then comprises a test step 250 during which an evolution of the observed scene is evaluated from the newly acquired image, in order to determine whether it is necessary to modify the integration time. This test step 250 may in particular be based on the calculation of an average filling rate of the detector wells in the image.

[0070] Where appropriate, the method comprises a step 252 of changing the current integration time, during which a new value of the current integration time is determined, for example to obtain a desired average filling rate.

[0071] This step 252 of changing the current integration time does not require going through a calibration step 210, and instead leads directly to a new acquisition step 230, with the new current integration time. The time difference calculated subsequently (236) will be different for this new image. interpolation table I will be automatically updated.

[0072] The method according to the invention therefore makes it possible to adjust the integration time dynamically, after each image if necessary, without having to go through a calibration of the camera each time which would make it unusable for a few seconds.

[0073] According to other embodiments, not shown in the figures, the function describing the variations of the offset of each pixel as a function of the integration time can be polynomial with a degree greater than 1, for example a polynomial of degree 2. It is then possible to measure several offset tables with distinct integration times, and to construct interpolation tables for each coefficient of the polynomial, in particular an interpolation table for each coefficient of the polynomial.

Claims

Claims

1. Method for using an infrared camera, comprising the following steps: - determining (214) a first reference integration time, - obtaining a first offset table associated with the first reference integration time, - determining (218) a second reference integration time different from the first reference integration time, - obtaining a second offset table associated with the second reference integration time, - constructing (222) at least one interpolation table representative of the variations of the offset as a function of the integration time from at least the first offset table and the second offset table, - acquiring (230) an image during a current integration time, and - correcting non-uniformities of the image acquired from the current integration time and the interpolation table.

2. Method according to the preceding claim, in which each interpolation table comprises values ​​of a coefficient of a polynomial function describing the variations of the offset as a function of the integration time for each pixel of the detector.

3. Method according to the preceding claim, in which the polynomial function is a linear function, and the interpolation table is unique and comprises leading coefficients.

4. Method according to the preceding claim, in which the step of correcting the non-uniformities of the image comprises a first step of correcting (234) the image by subtracting the first offset table and a step of adjusting (238) the corrected image by subtracting a product of the interpolation table by a time difference between the first reference integration time and the current integration time of said image.

5. Method according to one of the preceding claims, in which the first offset table, the second offset table and each interpolation table each comprise a distinct value for each pixel of the images acquired by the camera.

6. Method according to one of the preceding claims, comprising, after the step of correcting the non-uniformities of the image, a step of obtaining (240) a final image followed by a step of changing (252) the current integration time, the step of changing (252) being directly followed by a new step of acquiring (230) a new image with the new integration time.

7. Method according to the preceding claim, in which, during the changing step (252), the current integration time is modified to obtain a desired average filling rate of the pixels of the image.

8. Computer program comprising instructions for implementing the method according to one of claims 1 to 7, when said instructions are executed by a processor of a processing circuit.

9. A computer-readable recording medium comprising instructions which, when executed by a computer, cause the computer to carry out the steps of the method according to one of claims 1 to 7.

Images