Device and method for absolute calibration of imaging spectrometers
A compact calibration device with a broadband light source and diffuser system allows for efficient, reproducible absolute calibration of hyperspectral imaging systems, addressing the limitations of existing methods by enabling rapid and accurate calibration over large areas.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- INSTITUT PHOTOVOLTA QUE D ILE DE FRANCE (IPVF)
- Filing Date
- 2025-10-23
- Publication Date
- 2026-05-07
AI Technical Summary
Existing absolute calibration methods for hyperspectral imaging systems are time-consuming, prone to measurement errors, and lack reproducibility, especially when dealing with large fields of view, and require bulky equipment like integrating spheres.
A compact calibration device using a broadband light source with a cosine-corrected transmission diffuser and a stabilized intensity, coupled via optical fibers, enables absolute calibration by acquiring spectral data cubes and interpolating them to generate a field-of-view cube with correction parameters per pixel.
Facilitates rapid and reproducible absolute calibration over large areas without the need for multiple steps or bulky equipment, reducing measurement errors and enabling macroscopic maps with improved accuracy.
Smart Images
Figure EP2025080683_07052026_PF_FP_ABST
Abstract
Description
[0001] Device and method for absolute calibration of spectro-imagers
[0002] TECHNICAL FIELD
[0003] The field of the invention is that of the calibration of spectro-imagers, such as hyperspectral imagers in particular used to carry out luminescence imaging measurements (for example to characterize photovoltaic cells or to study semiconductors such as light-emitting diodes) or fluorescence imaging measurements (in the field of biology for example).
[0004] PREVIOUS TECHNIQUE
[0005] Despite significant advances in photovoltaic materials, new materials are struggling to be economically competitive with silicon-based materials. This is primarily due to a lack of understanding and control of the non-uniformity of active layers, thus hindering the optimization of optoelectronic properties. To bring the next generations of solar cells to market, researchers need to be able to study the spatial variation of their materials' properties on a larger scale.
[0006] To meet this need, hyperspectral imaging provides electroluminescence (EL) and photoluminescence (PL) maps that allow for the rapid characterization of structural and physical properties of a photovoltaic material.
[0007] The principle of hyperspectral measurement is as follows: the light spectrum of the object being tested (light response to a light flux or an electric current in the case of a solar cell) passes through the imager so that its image can be analyzed. An image is composed of the sum of point sources (at least one point source) originating from different positions of the object as seen by the imager's lens. Each point source produces a collimated polychromatic beam that arrives at a specific angle of incidence on a spectral filtering system such as a volume Bragg grating.
[0008] A volume Bragg grating is a diffraction grating that periodically modulates the refractive index across the volume of a photosensitive material (e.g., photothermo-refractive glass). This volume acts as a transparent window for all wavelengths except for a narrow spectral band of a few nanometers, which is diffracted in another direction at an angle dependent on the wavelength. Only this diffracted beam reaches a vertical line of the imager detector, which detects a one-dimensional, monochromatic image. The adjacent wavelength is projected onto the adjacent line; there is a different wavelength on each line. The wavelength selected on each line is adjusted by changing the angle of incidence between the polychromatic source beam and the grating. A scanning sequence is then performed.
[0009] Once all wavelengths have been scanned, an image reconstruction process is used which provides a "hyperspectral cube", i.e. one image per wavelength and one spectrum per pixel.
[0010] It should be noted that there are alternatives to the spectral filtering system constituted by the volume Bragg network, for example tunable bandpass filters.
[0011] Implementing absolute calibration of a hyperspectral imaging system allows researchers to determine the absolute number of photons emitted by each point on a sample's surface at each wavelength. This enables researchers to obtain quantitative information instead of data expressed in arbitrary units, thus providing a better understanding of the sample's optical properties.
[0012] An absolute calibration process, mentioned for example in Delamarre, A., Lombez, L., & Guillemoles, JF (2012). Characterization of solar cells using electroluminescence and photoluminescence hyperspectral images. Journal of Photonics for Energy, 2(1), 027004-027004, comprises four steps:
[0013] 1) The first step is to perform calibration with a line lamp. This is a classic procedure which consists of imaging lines of known emission wavelength to deduce the spectral dispersion of the imaging system and correct any spectral aberrations.
[0014] 2) The second step consists of performing a relative spectral calibration of the transmittance of the imaging system over the entire field of view. To do this, an extended source, spatially homogeneous over the area to be calibrated, with a known spectrum and Lambertian emission geometry (uniform luminance at each emitted wavelength) is imaged; this is done using a device called an integrating sphere coupled to a halogen lamp.
[0015] An integrating sphere is a spherical cavity covered with a perfect Lambertian reflector, with at least one orifice for injecting light from a lamp and at least one orifice for emitting Lambertian radiation. An integrating sphere with a diameter of 10 cm is typically used to generate a Lambertian source with a diameter of a millimeter.
[0016] A correction factor is calculated for each pixel and each wavelength which allows the spectral output of the system to match the known spectrum of the source.
[0017] 3) The third step consists of performing an absolute power calibration of a few pixels. For this, a monochromatic source of limited surface area and known power, whose emission cone is smaller than that of the imaging lens (i.e., a laser beam exiting a fiber), is imaged. Since the power is known, an absolute calibration of the system at the wavelength of the monochromatic source is obtained, from which it is possible to extrapolate to the entire spectral range using the relative calibration performed in the second step.
[0018] 4) The fourth step consists of performing a spatial calibration using a target. The aim is to calculate the spatial correspondence between the image of the target and a defined area of the sample, that is, to determine the correspondence between a pixel of the camera and a unit of length. This correspondence is necessary for calculating the detected light intensity in absolute units.
[0019] This absolute calibration method has drawbacks. First, the time required to perform the absolute calibration is significant, on the order of an hour, and the numerous steps involved introduce measurement errors and lead to many calculations. Second, power calibration is unstable and therefore has poor reproducibility. Similarly, the fourth step introduces a significant error / uncertainty into the calibration process. Furthermore, in addition to being expensive, the integrating sphere is bulky (at least 10 cm in diameter), which may sometimes require disassembling and reassembling the sample holder. Finally, this method cannot perform absolute calibration when the field of view becomes large (on the order of a centimeter), as the integrating sphere would then need to be more than one meter in diameter.
[0020] DESCRIPTION OF THE INVENTION
[0021] The invention aims to enable the absolute calibration of a spectro-imaging system while overcoming one or both of the aforementioned drawbacks. To this end, the invention proposes a calibration device for a spectro-imaging system, comprising, for achieving Lambertian scattering, a broadband light source calibrated in intensity and a cosine-corrected transmission diffuser coupled to said light source.
[0022] Some preferred, but not exhaustive, aspects of this system are as follows:
[0023] - the diffuser is coupled to the light source via an optical fiber;
[0024] - it also includes a ray lamp suitable for coupling to the diffuser;
[0025] - the light source is also stabilized in intensity;
[0026] - the diffuser includes a diffusing glass of the bubble glass type.
[0027] The invention also relates to a method for the absolute calibration of a spectro-imaging system using such a device. This method comprises:
[0028] - obtaining at least one cube of spectral data acquired by a camera of the spectro-imaging system with the diffuser of said device positioned in a field of view of the camera;
[0029] - the determination of a calibration cube from at least one spectral data cube obtained and an emission spectrum of the light source of said device, the calibration cube comprising, for each of a plurality of wavelengths, a correction parameter per pixel imaged by the camera.
[0030] In one possible embodiment, said obtaining comprises obtaining a plurality of spectral data cubes, each for one of a plurality of locations of the diffuser in the camera's field of view, and said determining comprises: o interpolating the obtained spectral data cubes to the whole field of view so as to obtain a field of view cube which includes, for each wavelength of said plurality of wavelengths, an intensity value per pixel imaged by the camera; and o comparing the field of view cube to the emission spectrum of the light source.
[0031] BRIEF DESCRIPTION OF THE DRAWINGS
[0032] Other aspects, objectives, advantages, and features of the invention will become clearer upon reading the following detailed description of preferred embodiments thereof, given by way of non-limiting example, and made with reference to the accompanying drawings in which:
[0033] - Figure 1 is a diagram of a device according to the invention for calibrating a spectro-imaging system;
[0034] - Figure 2A and Figure 2B represent measurements, respectively at 600 nm and 1300 nm, of the intensity identifier of the light scattered by a device according to the invention;
[0035] - Figure 3 represents, from left to right, an emission spectrum of the calibrated light source, a spectrum measured by a hyperspectral imaging system and the response of the hyperspectral imaging system, or device function;
[0036] - Figure 4 shows different possible locations of the diffuser for wide-field calibration;
[0037] - Figure 5 represents a correction surface determined by interpolation of spectral measurements carried out for each of the different locations in Figure 4.
[0038] DETAILED DESCRIPTION OF SPECIFIC METHODS OF IMPLEMENTATION
[0039] With reference to Figure 1, the invention proposes a device 1 for calibrating a spectro-imaging system, for example, for the absolute calibration of a spectro-imaging system such as a multispectral or hyperspectral imaging system. This device comprises, for achieving Lambertian light scattering, a light source 2 and a diffuser 3 coupled to the light source. A spectro-imager is understood here as a device that measures the light spectrum of one or more spatially delimited areas of a light-emitting object. In cases where multiple areas are observed, several scenarios are possible: either the detector comprises several pixels (hyperspectral imager using a camera), or a single area moves by scanning the object (confocal imager equipped with a spectrometer such as a confocal microscope). Ultimately, the spectro-imager provides a one- or two-dimensional optical image of the object with spectral resolution.
[0040] Light source 2 is a broadband source with calibrated intensity. A broadband source means that light source 2 generates a range of wavelengths that covers a large portion of the electromagnetic spectrum, for example, from 250 nm to 1500 nm. A source with calibrated intensity means that the emission spectrum of light source 2 is known. This emission spectrum may have been measured using a spectrogonometer.
[0041] The light source 2 is preferably intensity-stabilized to ensure the accuracy and reproducibility of the absolute calibration with respect to the calibration obtained from a calibrated spectrogonometer or the previous method described above. This stable operation can be ensured by current control of the power supply to the source 2. The stability of the light source is, for example, guaranteed to + / - 5% over 200 hours.
[0042] Alternatively, the power of the light source 2 can be measured during calibration and then relatively before each use.
[0043] Light source 2 can be a halogen lamp, a xenon lamp or a deuterium lamp.
[0044] Diffuser 3 is a cosine-corrected transmission diffuser that ensures the Lambertianity (or orthotropy) of the scattered light, especially within the spectral range of interest and over an angle greater than the aperture angles of the objectives used by the hyperspectral imaging system. Figures 2A and 2B show measurements of the intensity identifier of the scattered light at 600 nm and 1300 nm, respectively. These measurements demonstrate near-perfect Lambertianity at several wavelengths. Diffuser 3 may include a diffusing glass of the "milk glass" type. Such a milk glass can be a diffuser made from pure, opaque synthetic silica glass, such as the Diffusil® model from Opsira.
[0045] The light source 2 and the diffuser 3 can be coupled via an optical fiber 4, for example a fiber from a bundle of optical fibers.
[0046] The luminance of light from diffuser 3 on the central area (typically 1mm at the center), reasonably homogeneous (verified by imaging) is measured for example using a calibrated spectro-goniometer or a calibrated hyperspectral imager with the previous method described above.
[0047] Device 1 according to the invention allows for absolute calibration of a spectro-imaging system, such as a hyperspectral imaging system, without resorting to the second, third, and fourth steps of the prior method described above. This device is also more compact than the prior assembly comprising an integrating sphere coupled to a halogen lamp.
[0048] In a possible embodiment shown in Figure 1, the device 1 may further include a line lamp 5 suitable for being coupled to the diffuser, for example via an optical fiber 6. This optical fiber 6 may form a second path of a fiber bundle also accommodating the fiber 4 coupling the light source 2 and the diffuser 3. This line lamp can be used to carry out the first step of the previous process described above.
[0049] The invention is not limited to the device described above and also extends to the use of this device for calibrating a spectro-imaging system, such as a multispectral or hyperspectral spectro-imaging system. A hyperspectral spectro-imaging system may, in particular, utilize a Bragg volume as previously mentioned or a tunable bandpass filter.
[0050] This method is implemented following the positioning of the diffuser 3 of the calibration device 1 according to the invention within the field of view of a camera of the spectro-imaging system and the acquisition by the camera of at least one spectral data cube. A spectral data cube is here a stack of monochromatic images of the diffuser, with one spectrum per pixel imaged by the camera. The cube thus has two spatial dimensions (x and y) and one spectral dimension (z), the face of the cube being a function of the spatial coordinates and the depth being a function of the wavelength.
[0051] The method comprises obtaining at least one spectral data cube acquired by the camera and determining a calibration cube from the at least one spectral data cube obtained and an emission spectrum of the light source 2 of the calibration device 1 according to the invention. The calibration cube comprises, for each of a plurality of wavelengths, a correction parameter per pixel imaged by the camera.
[0052] Figure 3 shows on the left the emission spectrum λ(X) of the calibrated light source 2 (or irradiance), and in the center a spectrum S(X) per unit time measured by a hyperspectral imaging system at one pixel of the image of the diffuser 3 of the calibration device according to the invention. The raw spectrum measured per unit time S(X), shown in the center of Figure 3, depends on the response of the hyperspectral imaging system according to the equation S(X) = λ(X) * R(X), where R(A) corresponds to the response of the hyperspectral imaging system, or device function. Since λ(X) is known, it is possible to derive R(X) according to Φ(A) = On the right in Figure 3, R(X) represents a slice of the cube of calibration according to the spectral dimension, associated with a pixel of the image of diffuser 3.
[0053] Assuming the mechanical and optical stability of the imaging system, a subsequent measurement from a single point in the field of view allows, through relative calibration, the generation of a new calibration dataset. This method is well-suited to modifications of the imaging system, such as the addition of a filter.
[0054] This calibration method can be adapted to perform absolute calibration over a large area, for example, on the order of 15 x 15 cm, and to enable the creation of macroscopic maps, also known as wide-field maps. To achieve this, several spectral data cubes are acquired by moving the diffuser throughout the field of view. Thus, obtaining at least one spectral data cube involves acquiring a plurality of spectral data cubes, each corresponding to one of several diffuser locations within the camera's field of view. Figure 4 illustrates, as an example, different AY locations that can be used successively to acquire a spectral data cube.
[0055] In this scenario, the determination of the calibration cube includes the interpolation (for example, a biharmonic spline interpolation coupled with a natural neighbors extrapolation) of the spectral data cubes obtained across the entire field of view so as to obtain a field-of-view cube which includes, for each wavelength of said plurality of wavelengths, an intensity value per pixel imaged by the camera.
[0056] Figure 5 illustrates the result of such an interpolation, for a given wavelength (in this case 1100 nm), across the entire field of view of the measurements taken at the different locations and represented by circles. The field-of-view cube thus contains a correction surface across the entire field of view, of the type shown in Figure 5, for each wavelength. The determination of the calibration cube continues with a comparison of the field-of-view cube to the emission spectrum of the light source, according to the equation mentioned earlier.
[0057] The invention also extends to a data processing unit comprising a processor configured to implement the method described above, and to a computer program product comprising instructions which, when the program is executed by a computer, cause the computer to implement this method.
Claims
DEMANDS 1. Device (1) for calibrating a spectro-imaging system, comprising, for achieving Lambertian light scattering, a broadband light source calibrated in intensity (2) and a cosine-corrected transmission diffuser (3) coupled to said light source.
2. Device according to claim 1, wherein the diffuser (3) is coupled to the light source (2) via an optical fiber (4).
3. Device according to any one of claims 1 and 2, further comprising a line lamp (5) suitable for being coupled to the diffuser (3).
4. Device according to any one of claims 1 to 3, wherein the light source (2) is further stabilized in intensity.
5. Device according to any one of claims 1 to 3, wherein the diffuser (3) comprises a diffusing glass of the bubble glass type.
6. A method for the absolute calibration of a spectro-imaging system using a device (1) according to any one of claims 1 to 5, comprising: - obtaining at least one cube of spectral data acquired by a camera of the spectro-imaging system with the diffuser (3) of said device positioned in a field of view of the camera; - the determination of a calibration cube from at least one spectral data cube obtained and an emission spectrum of the light source of said device, the calibration cube comprising, for each of a plurality of wavelengths, a correction parameter per pixel imaged by the camera.
7. A method according to claim 6, wherein: - said obtaining includes obtaining a plurality of spectral data cubes, each for one of a plurality of locations (AY) of the diffuser in the camera's field of view; and - said determination includes: interpolating the obtained spectral data cubes to the whole field of view so as to obtain a field of view cube which includes, for each wavelength of said plurality of wavelengths, an intensity value per pixel imaged by the camera; and comparing the field of view cube to the emission spectrum of the light source.
8. Data processing unit, comprising a processor configured to implement the method according to one of claims 6 and 7.
9. Product computer program comprising instructions which, when the program is executed by a computer, cause the computer to carry out the method according to one of claims 1 and 7.