Multi-single detector application-specific spectrometer

Abstract

A spectral measurement device (110) for measuring optical radiation (112) provided by at least one measurement object (118) with a spectral measurement based on at least one classification, said spectral measurement device (110) comprising: - at least one radiation source (116) configured to emit optical radiation (112) at least partially towards said measurement object (118), said optical radiation (112) being at least partially within a spectral range that is subject to a classification-based spectral measurement; - at least two photodetectors (120), each photodetector (120) comprising at least one pixel (121), each pixel (121) being an active pixel (123) or a dark pixel (124), the spectral measurement device (110) comprising at least two active pixels (123), the spectral measurement device (110) comprising at least one dark pixel (124), each active pixel (123) configured to generate at least one photodetector signal dependent on illumination of the active pixel (123), at least two of the active pixels (123) configured to detect optical radiation in at least partially different spectral ranges, each dark pixel (124) configured to generate at least one photodetector signal independent of illumination of the dark pixel (124); - at least one readout device (139) configured to measure said photodetector signal and to generate at least one item of spectral data; at least one evaluation device (140) comprising at least one processor (142) and at least one memory storage (144), said memory storage (144) comprising: at least one classification model including a set of distinct classes, each class referring to at least one spectral characteristic, said classification model configured to classify at least one item of input data into said class; at least one transfer function configured to transfer the items of spectral data to items of input data applicable to the classification model; and at least one evaluation device (140) configured to store the Equipped with. The evaluation unit (140) is configured to generate at least one item of measurement information by applying the classification model and the transfer function to an item of the spectral data.

Description

[Technical field]

[0001] The present invention relates to a spectral measurement device for measuring optical radiation with at least one classification-based spectral measurement, a method for measuring optical radiation by performing a classification-based spectral measurement, and various applications of the spectral measurement device. Such devices and methods can generally be used for various applications. For example, such devices and methods can be employed for surveillance or monitoring purposes, in particular for infrared detection, heat detection, flame detection, fire detection, smoke detection, pollution monitoring, monitoring of industrial processes, chemical processes, food processing processes, sorting processes such as plastic sorting or fiber sorting, etc. However, further types of applications are also possible. [Background technology]

[0002] Spectrometers are used to measure the interaction of any substance with energy, for example electromagnetic radiation. For qualitative applications, such as the classification of different samples, the data is sent to a processing unit, such as a computer, where mathematical calculations are performed on the spectral data. However, the need to evaluate the spectroscopic data with a separate device reduces the capabilities of the spectrometer for consumer use. Furthermore, especially miniaturized reflectance spectrometers tend to be quite expensive for consumer use.

[0003] There are relatively inexpensive miniature spectrometry solutions based on microelectromechanical systems (MEMS), such as those presented by Spectral Engines GmbH. However, they are limited to a certain wavelength range, for example 1550 nm to 1950 nm for the product Nirone Sensor S. This technical limitation severely limits their use in applications where different wavelength regions are measured. Furthermore, moving MEMS parts introduce additional variability in the system performance. Mechanical movements such as vibrations can distort the measurement results. This makes it disadvantageous to use such miniature spectrometry solutions in consumer products or industrial applications. Furthermore, spectrometer systems such as Consumer Physics, Inc.'s SCiO™ have discrete wavelength filters, but their photosensitive detectors are CMOS imagers and their wavelength range is very limited to only a part of the near infrared region from 740 nm to 1070 nm. Furthermore, due to drifts of the photosensitive detectors, all such spectrometer systems may require manual calibration and are not easily usable. These shortcomings limit the application of qualitative analysis in consumer applications such as smartphones or smart home appliances such as vacuum robots or washing machines. Thus, miniaturization of spectrometers leads to increased use in industrial applications. However, as pointed out, miniature spectrometers have not yet found a foothold in consumer applications because they are relatively expensive, data interpretation is complicated, and they are difficult to incorporate into such consumer applications. For example, Consumer Physics, Inc.'s SCiO™ and Stratio, Inc.'s LinkSquare provide ultra-compact handheld spectrometers for consumer use, but the data output from the system is in the form of a spectrum, so data interpretation requires interpretation by a trained user.

[0004] EP 3 136 270 A1 discloses an apparatus receiving information identifying the result of a spectroscopic measurement of an unknown sample. The apparatus is capable of performing a first classification of the unknown sample based on the result of the spectroscopic measurement and a global classification model. The apparatus is capable of generating a local classification model based on the first classification. The apparatus is capable of performing a second classification of the unknown sample based on the result of the spectroscopic measurement and the local classification model. The apparatus is capable of providing information identifying a class associated with the unknown sample based on performing the second classification. However, the apparatus is not miniaturized and is therefore less suitable for consumer applications.

[0005] US 10,241,095 B2 discloses a series of optical spectral sensors for gas and vapor measurement using a combination of a solid-state light source (LED or broadband) and a multi-element detector housed in an integrated package including interface optics and acquisition and processing electronics. Spectral selectivity is provided by a custom detector. The multi-component gas monitor system has no moving parts and the gas sample flows in a measurement chamber and interacts with a light beam generated from a light source such as a MEMS broadband IR light source or an LED matrix. The custom detector is configured with multi-wavelength detection to detect and measure the light beam passing through the sample in the measurement chamber. However, the device does not perform sample classification, which reduces the information content for the consumer. Summary of the Invention [Problem to be solved by the invention]

[0006] It is therefore desirable to provide an apparatus and method that addresses the above-mentioned technical problems of known apparatus and methods. In particular, it is an object of the present invention to provide a reliable, low-cost, consumer-friendly apparatus and method for performing spectroscopic measurements with integrated data analysis. [Means for solving the problem]

[0007] This problem is solved by the invention with the features of the independent claims. Advantageous embodiments, which can be realized independently or in any combination, are set out in the dependent claims and throughout the specification.

[0008] In a first aspect of the present invention, a spectrum measurement device is disclosed for measuring optical radiation provided by at least one measurement object with a spectrum measurement based on at least one classification. As used herein, the term "spectrum", including any grammatical variations, is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer to a partition of a range of optical wavelengths, without being limited thereto. A spectrum may be constituted by an optical signal defined by a signal wavelength and a corresponding signal intensity. In particular, a spectrum may include spectral information related to the measurement object, for example related to the type and / or composition of at least one material forming the measurement object, which information can be determined by recording at least one spectrum related to the measurement object.

[0009] Thus, the term "spectral measurement device" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer, without limitation, to any device configured to determine spectral information by recording at least one measurement of at least one signal intensity associated with at least one corresponding signal wavelength of optical radiation and by evaluating at least one detector signal related to the signal intensity. Specifically, the spectral measurement device may be, comprise, or be part of at least one miniaturized device. For example, the spectral measurement device may be, comprise, or be part of at least one handheld device. The term "handheld" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer, without limitation, to any portable device. The handheld device may specifically be configured to be carried by a user in one hand due to its dimensions and / or its mass. Thus, as an example, the volume of a handheld device is 0.001 m 3 and / or the mass of the handheld device may not exceed 1 kg. In particular, the spectrum measurement device may be part of, comprise or be part of at least one wearable device, in particular a smartphone or a smartwatch. The fact that the spectrum measurement device may be part of, comprise or be part of a miniaturized device, such as a handheld device, may facilitate in particular a consumer-friendly application of the spectrum measurement device. Furthermore, the spectrum measurement device may comprise at least one housing. The housing may be configured to protect and / or shield the components inside the housing from environmental influences, such as mechanical or electromagnetic influences.

[0010] The term "optical radiation" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to any special or customized meaning. The term may specifically refer to, but is not limited to, one or more of the following: the visible spectrum range, the ultraviolet spectrum range, and the infrared spectrum range. The term "ultraviolet spectrum range" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to any special or customized meaning. The term may specifically refer to, but is not limited to, electromagnetic radiation having a wavelength of 1 nm to 380 nm, preferably 100 nm to 380 nm. Furthermore, in part in accordance with the version of standard ISO-21348 in effect at the date of this document, the term "visible spectrum range" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to any special or customized meaning. The term may specifically refer to, but is not limited to, the spectral range of 380 nm to 760 nm. Furthermore, the term "infrared spectral range" (IR) is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to any special or customized meaning. This term may specifically refer, without limitation, to electromagnetic radiation generally between 760 nm and 1000 μm, where the range between 760 nm and 1.5 μm is usually referred to as the "near infrared spectral range" (NIR), while the range between 1.5 μm and 15 μm is denoted as the "mid infrared spectral range" (MidIR) and the range between 15 μm and 1000 μm is denoted as the "far infrared spectral range" (FIR). Preferably, the optical radiation used for the general purpose of the present invention is optical radiation in the infrared (IR) spectral range, more preferably optical radiation in the near infrared (NIR) and mid infrared spectral range (MidIR), in particular optical radiation having a wavelength between 1 μm and 5 μm, preferably between 1 μm and 3 μm.

[0011] As mentioned above, the optical radiation is provided by at least one measurement object. In this context, the term "providing", as used herein and including any grammatical variations thereof, is a broad term and should be given its ordinary and customary meaning to a person skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer to at least one of reflecting, specifically reflecting diffusely; diffracting, specifically transmitting, and emitting optical radiation, specifically, without being limited thereto. The optical radiation provided by the measurement object may indicate at least one of a physical property of the measurement object, such as an optical property of the measurement object, and / or a temperature and a chemical property of the measurement object, such as a chemical composition of the measurement object. As an example, the optical radiation provided by the measurement object may be emitted by the measurement object, specifically emitted at least partially towards a spectral measurement device. Furthermore, the optical radiation provided by the measurement object may be reflected by the measurement object, specifically reflected at least partially towards a spectral measurement device, specifically reflected diffusely. Furthermore, the optical radiation provided by the measurement object may be transmitted at least partially through the measurement object towards the spectral measurement device. However, the measurement object may also at least partially absorb the optical radiation, which may in particular indicate at least one physical property of the measurement object and / or at least one chemical property of the measurement object, such as the chemical composition of at least one material forming the measurement object.

[0012] The term "measurement object" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer to anything selected from living and non-living objects, without being limited thereto. The measurement object may specifically include at least one material that is the subject of investigation by a spectrum measuring device. The measurement object may generally refer to an object to be measured, for example an object for which a spectrum is recorded, where the object has in principle any property, for example any optical property or any shape. The measurement object may specifically include at least one solid sample. However, other measurement objects, such as fluids, are also possible.

[0013] As mentioned above, the spectral measurement device is configured to measure the optical radiation with a spectral measurement based on at least one classification. The term "class" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer to a group of entities, such as objects or elements, that share at least one characteristic, without being limited thereto. The shared characteristic may refer to a physical characteristic, specifically an optical characteristic, such as a spectral characteristic, of the object or element. The shared characteristic may further refer to a chemical characteristic, such as a chemical composition of the object. As an example, the shared characteristic of the objects may be a common material that they all contain. Specifically, each class may refer to at least one material having at least one distinct spectral characteristic, or at least one group of materials having similar spectral characteristics. The spectral characteristic may specifically include at least one absorption peak of the material and / or a physical quantity related to the absorption peak. More specifically, the material may be at least one material selected from the group consisting of synthetic fibers, cotton, wool, silk, polyethylene terephthalate (PET), polypropylene (PP), polyethylene (PE), polyvinyl chloride (PVC), high density polyethylene (HDPE), low density polyethylene (LDPE), polyamide (PA), and glass. In general, the term category may be used synonymously with the term class.

[0014] Thus, the term "classification" as used herein, including grammatical variations, is a broad term and should be given its ordinary and customary meaning to those skilled in the art, and should not be limited to any special or customized meaning. The term may specifically, but without limitation, refer to grouping entities having shared properties within their respective classes. Specifically, multiple entities having at least one shared property can be grouped into one class. Classification may include first analyzing the entities and identifying the shared properties. As an example, classification may specifically refer to analyzing the chemical composition of the objects and grouping objects containing a common material into one class. Classification may also refer to the creation and / or definition of classes, specifically the definition of class boundaries. As an example, classification may set at least one limit of a physical property, such as at least one of density, concentration, and concentration, as a class boundary. As a further example, classification may set at least one maximum deviation from a predetermined condition or state as a class boundary. For example, a measurement object consisting mainly of one material except for a predetermined deviation portion can be grouped into one class. Thus, classes can be specifically defined by material. Classification may include assigning an entity to at least one class. In principle, it is also possible to assign an entity to several classes simultaneously. Thus, given a defined class, at least one entity can be assigned to the defined class, specifically if it shares with other entities at least one characteristic that the class already contains, or if it is within the boundaries of the defined class. As an example, classes can be defined by material. Then, the measured object can be assigned to a class or category if it contains the material at least to a predetermined extent (which can be determined by measuring at least one physical property of the object, such as, for example, an optical property, e.g., absorption).

[0015] The term "classification-based spectral measurement" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to any special or customized meaning. The term may specifically refer, without limitation, to a spectral measurement in which measured spectral data and / or pre-processed measured spectral data are subsequently classified. The result or intermediate result of the classification-based spectral measurement may specifically include information regarding the class to which the measurement object is assigned.

[0016] Spectral measurement equipment: - at least one radiation source configured to emit optical radiation at least partially towards the measurement object, the optical radiation being at least partially within a spectral range that is subject to a classification-based spectral measurement; - at least two photodetectors, each photodetector comprising at least one pixel, each pixel being an active pixel or a dark pixel, said spectral measurement device comprising at least two active pixels, said spectral measurement device comprising at least one dark pixel, each active pixel configured to generate at least one photodetector signal dependent on illumination of said active pixel, at least two of said active pixels configured to detect optical radiation in at least partially different spectral ranges, each dark pixel configured to generate at least one photodetector signal independent of illumination of said dark pixel; - at least one readout device configured to measure said photodetector signal and generate at least one item of spectral data; at least one evaluation device comprising at least one processor and at least one memory storage, said memory storage comprising: at least one classification model including a set of distinct classes, each class referring to at least one spectral characteristic, said classification model configured to classify at least one item of input data into said class; at least one transfer function configured to transfer the items of spectral data to items of input data applicable to the classification model; At least one evaluation device configured to store Equipped with.

[0017] The evaluation unit is configured to generate at least one item of measurement information by applying the classification model and the transfer function to an item of spectral data.

[0018] The term "radiation source" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer to, without limitation, any device configured to emit optical radiation in principle. The term "emit" as used herein, including its grammatical variations, is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer to, without limitation, any process that generates and transmits optical radiation. The radiation source may comprise at least one of a semiconductor-based radiation source and a thermal emitter. The semiconductor-based radiation source may be selected from at least one of a light-emitting diode (LED) or a laser, specifically a laser diode. The thermal emitter may comprise at least one incandescent lamp.

[0019] The radiation source may be modulated. The term "modulate" as used herein, including its grammatical variations, is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer to, but is not limited to, the process of varying, specifically periodically varying, at least one characteristic of the optical radiation, specifically one or both of the intensity or phase of the optical radiation. The modulation may be a full modulation from a maximum value to zero, or a partial modulation from a maximum value to an intermediate value greater than zero. The radiation source may be electrically and / or mechanically and / or electromechanically modulated. As an example, the radiation source may be modulated by using at least one chopper.

[0020] The term "spectral range subject to classification-based spectral measurement" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. This term may specifically refer to, without limitation, a spectral range suitable for measuring at least one optical property of a measurement object that can be used directly or indirectly to classify the measurement object. As an example, within the spectral range, the measurement object may exhibit at least one optical property characteristic of a material, such as a spectral characteristic, e.g., an absorption peak, which can be used to assign the measurement object to a class defined by the material.

[0021] The term "photodetector" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to any special or customized meaning. The term may specifically refer to, without limitation, an optical sensor configured to detect optical radiation, such as for detecting illumination and / or a light spot generated by at least one light beam. The photodetector may comprise at least one substrate. A single photodetector may be a substrate having at least one single light-sensitive area that generates a physical response to illumination of a given wavelength range. As mentioned above, each photodetector comprises at least one pixel. The term "pixel" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to any special or customized meaning. The term may specifically refer to, without limitation, a unit of light-sensitive area of ​​a photodetector. A pixel may be illuminated by optical radiation and may generate at least one photodetector signal in response to illumination. The photodetector signal may comprise any signal generated by a photodetector. The pixels may be arranged on the surface of the photodetector, in particular on a substrate, and may in particular be a single, closed, uniform light-sensitive area.

[0022] At least two of the pixels may be different in size. The sizes of the pixels may be different from each other. The pixel sizes from different substrates or even the pixel sizes of the dual pixel detector may be different. The size of each pixel may be such that at least a similar photodetector signal intensity is generated by each photodetector, taking into account the photosensitivity of each pixel at a given wavelength and the intensity of the optical radiation incident on each pixel. In this way, it is possible to obtain similar signal intensities from multiple detectors during a measurement, which may be useful for optimizing the readout electronics. Other embodiments of the pixels may also be possible.

[0023] The pixels of each photodetector can be manufactured from the same material. Alternatively, pixels of different pixel materials can be combined into a single detector assembly. In this way, the spectral measurement device can cover a wide wavelength range and can be optimized in terms of performance, for example by varying the size of the active area. As an example, the single detector assemblies can have PbS and PbSe pixels of different sizes on different substrates.

[0024] As mentioned above, each pixel is an active pixel or a dark pixel. The substrate may have only one photosensitive area or a dual photosensitive area, with one "active" area and one second "dark" area, by mechanical or optical means. The spectral measurement device comprises at least two active pixels. Each active pixel is configured to generate at least one photodetector signal dependent on illumination of the active pixel. At least two of the active pixels are configured to detect optical radiation at least partially in different spectral ranges. The term "active pixel" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer, without limitation, to a pixel that is arranged and / or configured to generate a photodetector signal in response to illumination of its photosensitive area. Active may refer to the photosensitive area being illuminated in a sensitivity wavelength range during measurement. The active pixel may generate a photodetector signal dependent on illumination of the active pixel.

[0025] The spectral measurement device comprises at least one dark pixel. The at least two active pixels and at least one dark pixel may be distributed arbitrarily on the at least two photodetectors. As an example, the spectral measurement device may comprise two photodetectors, one photodetector comprises one active pixel and one dark pixel, and one photodetector comprises one active pixel. The term "dark pixel" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer to a covered pixel, without being limited thereto. Dark may refer to no radiation in the sensitivity wavelength range of the photosensitive area reaching the photosensitive area. The dark pixel may be covered by a material that absorbs in the same wavelength range as the sensitivity range of the photodetector and / or it may be a metal piece on the photosensitive area that reflects incident light away from the dark pixel. The dark pixel may be covered with a material that absorbs light radiation, such as glue and / or ink. Additionally or alternatively, the dark pixels may be covered with at least one material that reflects optical radiation, such as a metal. The dark pixels may in particular be completely invisible or at least barely visible to optical radiation. The dark pixels may be embodied as additional photodetectors with a single pixel covered by the above-mentioned means, or photodetectors with dual pixels (on one single substrate) may be used, where one of the pixels is darkened. The dark pixels may be used to eliminate long-term drift and temperature effects, as will be explained in more detail below.

[0026] The dark pixel is configured to generate at least one photodetector signal that is independent of illumination of the dark pixel. The term "illumination independent" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically, but without limitation, refer to the fact that the dark pixel may be configured to generate a photodetector signal without photons incident on the dark pixel, particularly photons in a wavelength range to which the dark pixel may be sensitive. In particular, the term "illumination independent" may refer to the fact that illumination of the dark pixel may not affect the photodetector signal generated by the dark pixel. As an example, high-intensity illumination of the dark pixel may result in the same photodetector signal as low-intensity illumination of the dark pixel. Illumination of the dark pixel in a different wavelength range may result in the same photodetector signal. The dark pixel may be configured to generate a photodetector signal without being illuminated. Thus, the photodetector signal generated by the dark pixel may depend at least primarily on the properties of the dark pixel itself, such as the material used in the dark pixel. The photodetector signal generated by the dark pixel may specifically include a dark signal, more specifically a dark current.

[0027] Each photodetector may include up to one active pixel, and the spectral measurement device may include at least one dark pixel. For example, each of the photodetectors may include two pixels, an active pixel and a dark pixel. For example, one of the photodetectors may include a single active pixel and another photodetector may include a dark pixel. No photodetector may have more than one active pixel. Each active pixel may be configured to detect a specific spectral range and / or wavelength range. Thus, each photodetector may be responsible for only a specific spectral range.

[0028] As an example, the spectral measurement device may include three pixels, two of which are active pixels and one of which is a dark pixel. Each photodetector may include at least one pixel, but not more than one active pixel. A photodetector with at least two pixels may have all but one pixel darkened to have only one active pixel. Each active pixel may be responsible for only one wavelength range, and there cannot be two active pixels for the same wavelength range in the spectral measurement device. The spectral measurement device may include, for example, two photodetectors, where a first photodetector may include two pixels, one active pixel and one dark pixel, and a second photodetector may include one active pixel. Furthermore, the spectral measurement device may include three photodetectors, each with one pixel, where a first photodetector may include a dark pixel, and a second and third photodetector may each include an active pixel.

[0029] At least one of the photodetectors may comprise at least two pixels. At least one of the pixels may be an active pixel. At least one of the pixels may be a dark pixel. In particular, the photodetector may be a dual pixel photodetector. Thus, the photodetector may comprise an active pixel and a dark pixel. The pixels may comprise the same photosensitive material. Thus, the calibration performed by using the dark pixels may be based on the same photosensitive material as the active pixels comprise. All pixels on one photodetector may comprise the same photosensitive material.

[0030] The spectral measurement device may comprise at least two radiation sources configured to emit light radiation in at least partially different spectral ranges. As an example, the spectral measurement device may comprise two different radiation sources, for example two different LEDs. Additionally or alternatively, the two radiation sources may comprise different light filters. Each photodetector may be configured to detect light radiation emitted by a different and exactly one radiation source. Each photodetector may be assigned to observe a different spectral range. Each photodetector may be assigned to observe a spectral characteristic corresponding to a different class of the classification model. The spectral measurement device may comprise at least three photodetectors, in particular at least four photodetectors, more particularly at least five photodetectors. For example, the spectral measurement device may comprise 16 or fewer photodetectors. In particular, the spectral measurement device may comprise at least three pixels. More particularly, the spectral measurement device may comprise at least two active pixels and at least one dark pixel. Each active pixel may be provided by a different photodetector. The two active pixels may each be used to observe a different spectral range. The dark pixels can be used to recalibrate the spectral measurement device, in particular to continuously or repeatedly recalibrate the spectral measurement device, for example in parallel with performing a classification-based spectroscopic measurement.

[0031] The evaluation device may be configured to automatically recalibrate the spectral measurement device by using at least one photodetector signal generated by using the dark pixels. The term "automatic" as used herein, including its grammatical variations, is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically, without limitation, refer to a process that is at least partially performed without requiring human intervention, e.g., at least partially by a machine. In particular, the process may be at least partially performed by at least one of a processor, a controller, a computer, a computer network, and a machine, in particular without manual actuation and / or user interaction. The term "calibrate" as used herein, including its grammatical variations, and including the term "recalibrate" is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer to, but is not limited to, correcting drift effects that may occur in the actual measurement due to changes primarily related to hardware components and / or due to changes affecting the hardware components, specifically the spectral measurement device or parts thereof. The changes may specifically include at least one of the following: degradation of the radiation source or photodetector, specifically the pixels; temperature drift of at least one of the radiation source or photodetector; variations in the ambient temperature affecting the spectral measurement device; variations in temperature related to the spectral measurement device, i.e., variations in the temperature at which the photodetector and / or the corresponding electronics may operate; mechanical expansion and contraction of at least one component contained by the spectral measurement device, specifically the mechanical housing, holder and / or optical elements. Further changes may also be possible. Physical processes such as electrochemical processes or relaxation of long-lived traps may induce drift effects.Correcting drift effects can facilitate maintaining the reliability of an item of measurement information, in particular by avoiding that drift effects distort the item of measurement information to such an extent that the results determined by the spectral measurement device become inconclusive. Periodic calibration of the hardware used may therefore typically be necessary to maintain the reliability of the spectral measurement device. Automatically calibrating the spectral measurement device by using at least one photodetector signal generated by using a dark pixel may be particularly consumer-friendly. The photodetector signal generated by using a dark pixel may be suitable for correcting photodetector drifts, in particular long-term drifts, and also for correcting temperature effects on the photodetector.

[0032] Each photodetector, in particular its pixel, is configured to generate at least one photodetector signal, also denoted detector signal, depending on the illumination of the pixel. The photodetector signal may comprise at least one of an analog signal and a digital signal. The photodetector signal may in particular comprise at least one electronic signal related to the intensity of the optical radiation incident on the pixel. At least one of the photodetectors may comprise at least one signal processing device, for example one or more filters and / or an analog-to-digital converter for processing and / or pre-processing the photodetector signal.

[0033] The photodetector signal generated by using dark pixels, also called dark signal, may be independent of the optical radiation reaching the spectral measurement device. In other words, the photodetector signal generated by using dark pixels may be constant even if the optical radiation reaching the spectral measurement device changes. The photodetector signal generated by using dark pixels may be suitable for correcting the drift of the photodetector, especially the drift over time. The photodetector signal generated by using dark pixels may be temperature dependent. As an example, the dark pixels may include at least a semiconductor material in which the spontaneous formation of free charge carriers can be thermally induced. Thus, the photodetector signal generated by using dark pixels may be even more suitable for correcting the temperature effect on the photodetector.

[0034] At least one of the photodetectors, specifically at least one of the pixels, may comprise a photoconductive material, specifically an inorganic photoconductive material selected from lead sulfide (PbS), lead selenide (PbSe), germanium (Ge), indium gallium arsenide (InGaAs, including but not limited to extended InGaAs), indium antimony (InSb), or mercury cadmium telluride (HgCdTe or MCT). The commonly used term "extended InGaAs" refers to a specific type of InGaAs that exhibits a spectral response up to 2.6 μm. Specifically, at least one of the photodetectors may be a PbS photodetector, with the emitted optical radiation including wavelengths between 760 400 nm and 3000 nm, specifically between 1000 800 nm and 2700 nm, more specifically between 1500 1200 nm and 2500 nm. However, different types of materials or other types of photodetectors may also be possible.

[0035] The photodetectors may be separated from one another in a manner that suppresses crosstalk between the individual detectors and stray light. This is possible because the photodetectors are physically separated on their respective substrates. Thus, radiation at a specified wavelength can be detected more accurately without crosstalk.

[0036] At least one of the radiation sources and / or photodetectors may be positioned such that directly reflected optical radiation from the measurement object does not reach the active pixels. The photodetectors may be configured to measure diffusely reflected optical radiation from the measurement object. Each of the photodetectors may be configured to measure diffusely reflected light from the measurement object and generate a photodetector signal. Directly reflected light from the measurement object can be prevented as far as possible from reaching the photodetectors by a suitable positional relationship between the at least one radiation source and the photodetectors.

[0037] As mentioned above, at least two of the photodetectors are configured to detect optical radiation in at least partially different spectral ranges. The term "at least partially different spectral ranges" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer to non-identical spectral ranges, without limitation, and the spectral ranges may or may not overlap. At least two of the photodetectors may be different, for example, including pixels including different photoconductive materials. Specifically, each active pixel may be configured to detect optical radiation having at least one spectral characteristic that refers to at least one class of the classification model. Each active pixel may be configured to detect optical radiation having at least one spectral characteristic that refers to exactly one class of the classification model. As mentioned above, each class may specifically refer to at least one material having at least one distinct spectral characteristic, or at least one group of materials having similar spectral characteristics. Each spectral characteristic may refer to at least one predetermined and / or predefined spectral range. Spectral characteristics, e.g., absorption peaks of a material, may be located in a predetermined and / or predefined spectral range. The predetermined and / or predefined spectral range may be predetermined and / or predefined before measuring the optical radiation provided by the measurement object, in particular according to a predetermined classification. As an example, a classification-based spectral measurement may be envisaged for determining materials in a garment. Before measuring the optical radiation provided by the garment, fiber classes may be predetermined and / or predefined, each fiber class may include spectral characteristics, e.g., absorption peaks, in different predetermined and / or predefined spectral ranges.The predetermined and / or predefined spectral range may be at least one spectral range selected from the group consisting of: 1650 nm to 1700 nm; 1725 nm to 1800 nm; 2050 nm to 2150 nm; 2175 nm to 2225 nm, however, other predetermined and / or predefined spectral ranges are possible for other use cases.

[0038] At least one of the optical detectors may comprise at least one optical filter. For example, the optical filter may be an optical bandpass filter. For example, the optical filter may be a narrow bandpass (NPB) optical filter. The spectral range detected by the optical detector may be limited to a spectral range associated with at least one class of the classification, e.g., exactly one class of the classification. The optical detector may be assigned to observe optical characteristics associated with at least one class of the classification, specifically, exactly one class of the classification. Each optical detector of the spectral measurement device may be assigned to observe a different spectral range. Each optical detector may be assigned to observe spectral characteristics corresponding to a different class of the classification model. The term "optical filter" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer to any device or object configured to selectively block optical radiation according to at least one physical characteristic of the optical radiation, such as, but not limited to, the wavelength of the optical radiation and / or the polarization of the optical radiation. The term "optical bandpass filter" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to any special or customized meaning. The term may specifically refer, without limitation, to an optical filter configured to filter incident optical radiation such that only a limited spectral range can pass through the optical filter and propagate to an object behind the optical filter. An optical bandpass filter may be configured to transmit only optical radiation from a minimum wavelength to a maximum wavelength.

[0039] The spectral measurement device can be designed specifically for each use case. The spectral measurement device can be configured to measure application-specific absorption bands. The photodetectors can be configured to the absorption bands and can include optical filters, such as narrow band pass (NBP) optical filters, with application-specific center wavelengths, full width at half maximum (FWHM), out-of-band cutoff levels and ranges. Additionally or alternatively, the photodetectors can be configured to those application-specific absorption bands by selecting appropriate radiation sources, such as LEDs, that emit sample-specific wavelengths.

[0040] For example, a textile use case may call for differentiation of different fibers into classes, e.g. cotton, wool, silk, synthetics, etc. All fibers have a certain characteristic infrared absorption spectrum. It has already been shown in the literature that certain materials have specific absorption bands. For example, cotton absorbs around 2100 nm, wool absorbs around 1724 nm, synthetics around 1662 nm, etc. Therefore, in this textile use case, instead of measuring the entire spectrum with an expensive spectrometer, it may be sufficient to measure these four material-specific absorption bands by setting the detector pixels to these sample-specific wavelengths, since only certain wavelength components carry information about the fiber class. Four photodetectors can be used, e.g. photodetectors 1 to 4. The photodetectors can be configured for the wavelengths using narrow band pass (NBP) optical filters with center wavelengths and full width at half maximum (FWHM) values ​​as shown in the table below: [Table 1]

[0041] For example, a further use case may require classification of plastics, e.g., PET is classified from other plastic types such as PP, PE, etc. In this embodiment, two or three pixels specific to these plastic classes can be used.

[0042] A hardware solution designed for one use case may not be applicable to another use case. Application specific designs proposed herein may include predefined hardware and classification models for a particular use case.

[0043] The spectrum measuring device comprises at least one readout device. The term "readout device" as used herein is a broad term and should be given its usual and customary meaning to a person skilled in the art and should not be limited to a special or customized meaning. The term may refer in particular, but not limited to, any device configured to quantify and / or process at least one physical property and / or changes in at least one physical property detected by at least one photodetector, in particular by at least one photodetector. As mentioned above, the at least one photodetector may in particular comprise at least one photoconductive material, such as, for example, PbS. Photoconductive materials generally undergo a change in conductivity upon irradiation and therefore a change in resistance, which can be quantified, for example, in a resistance measurement. The measured resistance value allows conclusions to be drawn regarding the incident light radiation. The resistance measurement may be performed in comparison with a further resistor, in particular with a further resistor having a known resistance value. The further resistor may in particular be arranged in a voltage divider in the readout device. The readout device may comprise at least one of an ohmmeter, a voltmeter, an ammeter, a lock-in amplifier, a voltage divider, and an electrical pass filter.

[0044] The readout device is configured to measure the photodetector signals and generate at least one item of spectral data. The term "item of spectral data" as used herein is a broad term and should be given its ordinary and customary meaning to a person skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer to at least one of knowledge or evidence providing a qualitative and / or quantitative description of spectral information, e.g., spectral characteristics, of a measurement object, without being limited thereto. The item of spectral data may relate to the absorption of optical radiation by the measurement object. The item of spectral data may include at least one absorption spectrum of the measurement object, the absorption spectrum may include at least one absorption peak. The item of spectral data may be analog and / or digital. The readout device may be configured to collect detector signals, each detector signal may, for example, be indicative of a detected intensity of optical radiation in a different spectral range. The readout device may be configured to process and / or pre-process the detector signals. The readout device may be configured to combine the detector signals in at least one spectrum, specifically at least one absorption spectrum of the measurement object.

[0045] The spectrum measuring device comprises at least one evaluation device. The term "evaluation device" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer to any device adapted to perform specified operations, preferably by using at least one data processing device, more preferably by using at least one processor and / or at least one application specific integrated circuit, without being limited thereto. As an example, the at least one evaluation device may comprise at least one data processing device on which software code including a number of computer commands is stored. The evaluation device may be provided with one or more hardware elements for performing one or more of the specified operations and / or one or more processors having software running thereon for performing one or more of the specified operations. As an example, the evaluation device may comprise one or more computers, application specific integrated circuits (ASICs), digital signal processors (DSPs), or one or more programmable devices, such as field programmable gate arrays (FPGAs), configured to perform the evaluation. However, additionally or alternatively, the evaluation device may also be fully or partially embodied by hardware. The evaluation device may further be configured to control the spectral measurement device or a part thereof. The evaluation device may in particular be configured to perform at least one measurement cycle during which a plurality of photodetector signals may be picked up. Information determined by the evaluation device may in particular be provided to at least one of the further devices and / or to a user in at least one of electronic, visual, acoustic or tactile manners. Information determined by the evaluation device may be stored in a memory storage and / or in a separate storage device and / or may be communicated via at least one interface, such as a wireless interface and / or a wired interface.

[0046] The evaluation device comprises at least one processor. The term "processor" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to any special or customized meaning. The term may refer specifically, but not limited to, any logic circuit configured to perform the basic operations of a computer or system, and / or generally, a device configured to perform calculations or logical operations. In particular, the processor may be configured to process basic instructions that run the computer or system. As an example, the processor may comprise at least one arithmetic logic unit (ALU), at least one floating point unit (FPU), such as a math coprocessor or numeric coprocessor, a number of registers, specifically registers configured to provide operands to the ALU and store operation results, and memories, such as an L1 cache memory and an L2 cache memory. In particular, the processor may be a multi-core processor. In particular, the processor may be or comprise a central processing unit (CPU). Additionally or alternatively, the processor may be or comprise a microprocessor, and thus, in particular, the elements of the processor may be included on one single integrated circuit (IC) chip. Additionally or alternatively, the processor may be or comprise one or more application specific integrated circuits (ASICs) and / or one or more field programmable gate arrays (FPGAs), etc. The processor may be configured, in particular, by software programming, etc., to perform one or more evaluation operations.

[0047] The evaluation device comprises at least one memory storage. The term "memory storage" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may refer, in particular, but not limited to, any device configured to store data, in particular to store data in an organized manner, in particular in a database, more particularly in at least one database record. The memory storage may be electronic and / or magnetic and / or mechanical memory storage. The memory storage may be writable or read-only. The memory storage may be volatile or non-volatile. The memory storage may comprise at least one of a random access memory (RAM), a read-only memory (ROM) and a flash memory. The memory storage may comprise at least one of a hard disk drive (HDD), a solid-state drive (SSD) and a flash drive.

[0048] The memory storage is configured to store at least one classification model. The classification model comprises a set of different classes. Each class refers to at least one spectral characteristic. The classification model is configured to classify at least one item of the input data into a class. The term "classification model" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer to a model or scheme that can be used to perform at least one classification, without being limited thereto. The classification model may include pre-determined and / or pre-defined classes, such as pre-determined and / or pre-defined classes by a user. The classification model may include class boundaries. The classification model may include class properties, such as at least one property shared by all entities in a class. The classification model is capable of classifying items of the input data into exactly one class. The classification model is capable of classifying items of the input data into multiple classes.

[0049] The classification model may be a global classification model. The term "global classification model" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer to, but is not limited to, a hardware-independent but application-specific classification model. The classification model may depend on use-case specific wavelengths, particularly spectral ranges. There may be a unique global model for every application. The classification model may be independent of hardware conditions and / or operating conditions. Thus, the classification model can be applied globally, particularly to different hardware components, such as photodetectors used under different operating conditions, such as temperature.

[0050] A classification model may include at least one algorithm for analyzing items of input data and classifying items of input data into classes. A classification model may include at least one trained model. A trained model is trained with at least one training set, specifically by using machine learning. In this context, the term "training" as used herein, including its grammatical variations, is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer to, without limitation, a process of determining parameters of a model and / or parameters used within an algorithm by using at least one training data set. Specifically, an algorithm may include a model. Training may include at least one optimization or tuning process, where an optimal parameter combination may be determined. As shown, training may be performed by using machine learning. The term "machine learning" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer, without being limited thereto, to methods that use artificial intelligence (AI) for automatic model building, especially for parameterizing models.

[0051] The training data set may for example comprise past spectral data acquired by at least one external device, in particular by at least one further spectral measurement device, which may be a laboratory spectral measurement device.

[0052] The classification model can be generated by using at least one of random forest, K-means clustering, and support vector machine (SVM), depending on the application and the subsequent accuracy. The term "random forest" (also written as random forest algorithm) as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art, and should not be limited to any special or customized meaning. The term may specifically refer to an ensemble learning method configured for at least classification and / or regression, without being limited thereto. In particular, the random forest algorithm may be configured to build one or more decision trees and output at least one class selected from the mode of the class and the average predicted value of the individual trees. The random forest algorithm is generally known, for example, from the academic paper "Random Forests" by Leo Breiman (Machine Learning 45.1, October 2001 issue). The term "K-means clustering" (also written as K-means algorithm) as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art, and should not be limited to any special or customized meaning. The term may specifically, but is not limited to, refer to algorithms that use vector quantization to classify entities. K-Means clustering may be performed as described in JA Hartigan, MA Wong: "Algorithm AS 136: A K-Means Clustering Algorithm", Journal of the Royal Statistical Society, Series C (Applied Statistics). Vol. 28, No., 1979, pp. 100-108.The term "SVM" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to any special or customized meaning. The term may specifically refer to a set of related supervised learning methods used for classification and regression, without limitation, see, for example, V. Vapnik. "The Nature of Statistical Learning Theory", Springer, NY, 1995. ISBN 0-387-94559-8.

[0053] The classification model is configured to classify at least one item of input data into a class. The term "item of input data" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer to at least one of knowledge or evidence providing a qualitative and / or quantitative description related to information, recorded and / or structured in such a way that the information can be interpreted within at least one entity, e.g., a model, a program, a processor, or a network, in particular for further processing, without being limited thereto. In particular, the item of input data is applicable to the classification model as described above. The classification model may be configured to interpret the item of input data, to process the item of input data, in particular to classify the item of input data and / or the item of processed input data. The classification model may be particularly application-oriented. As an example, the classification may refer to a material analysis of clothing, as described above, and the classes may be fiber classes, such as cotton or wool. In this example, the classification model may be configured to classify input data in the form of processed spectral data referring to clothing into fiber classes.

[0054] The memory storage is configured to store at least one transfer function. The transfer function is configured to transfer the item of spectral data to an item of input data applicable to the classification model. The global classification model can be adopted to a similar hardware designed for that particular use case with the help of the transfer function. The data measured by the spectral measurement device strongly depends on the tolerances of the hardware as well as on the ambient conditions such as temperature and humidity. As an example, the readout device of the spectral measurement device can generate an item of spectral data by a spectral measurement of a measurement object. The item of spectral data may first depend on the measurement object, e.g., the absorption characteristics of the measurement object. However, the item of spectral data may further depend on the spectral measurement device itself, in particular the hardware of the spectral measurement device, such as the photodetector used. Furthermore, the item of spectral data may depend on the operating conditions (e.g., temperature) of the spectral measurement device during the spectral measurement. The transfer function can transform the data measured by the spectral measurement device such that the data can be applied to the global classification model offline on a similar device. A similar device may refer to a predefined hardware having a predefined filter, number of pixels, pixel shape, etc.

[0055] The term "transfer function" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may refer in particular, without limitation, to any function or algorithm for transferring data. The transferring may include at least one transformation of the data, in particular adapting the data for further processing, more particularly adapting the data for further processing within at least one particular entity, such as a particular model. The transferring may include transferring the data in such a way that the information contained by the data is at least partially maintained. In particular, information relevant to further evaluation of the data is maintained, but the format of the data may be changed. Additionally or alternatively, further information may be added. For example, information regarding the hardware and / or operating conditions may be added to an item of spectral data to take into account the hardware and / or operating conditions for further evaluation, and / or the item of spectral data may be adapted by using a transfer function, in particular by taking into account information regarding the hardware and / or operating conditions when generating at least one item of input data applicable to the classification model. The transfer function may include at least one parameterized mathematical function including at least one parameter. The transfer function may be a function of coefficients corresponding to the pixels. The transfer function may take into account at least one of the hardware of the spectral measurement device and the operating conditions of the spectral measurement device. In particular, the parameters of the transfer function may be selected to weight the photodetector signals, more particularly by at least one of the applied shape of the pixel and the applied material of the pixel. As an example, the spectrometer device may comprise at least two pixels having different shapes and / or different active materials. Thus, the two pixels may induce different photodetector signals for the same incident optical radiation containing the spectral information of one measurement object.By using a transfer function to transfer an item of spectral data to an item of input data applicable to the classification model, different shapes and / or different active materials may be taken into account when classifying the measured object, for example, into at least one material class, by using the classification model. The transfer function allows the material classification application to be localized in hardware, allowing for mass deployment. The transfer function may be derived from an item of known spectral data, for example obtained by using a reference device on a reference sample. The reference device may comprise, for example, a high-end lab device. The reference sample may comprise, for example, a known sample, such as a known material. In particular, the transfer function may include at least one pre-determined coefficient. The parameterized mathematical function may include at least one matrix M of: Y=MX, where Y is a vector containing items of input data and X is a vector containing items of spectral data. The predetermined coefficients may thus be, for example, elements of a matrix M. The parameters of the matrix M may be determined by comparing the items of measured spectral data with items of known spectral data, e.g. obtained by using a reference device on a reference sample. The matrix M may, for example, contain device characteristics, e.g. detection rate, or temperature or humidity compensation parameters. The transfer function may further comprise at least one look-up table. The look-up table may be configured to match items of spectral data with items of input data applicable to the classification model.

[0056] As a particular example, the transfer function may allow for temperature compensation, so that the matrix M may include temperature compensation parameters as elements, which may be retrieved, for example, from a look-up table.

[0057] The evaluation device is configured to generate at least one item of measurement information by applying the classification model and the transfer function to the item of spectral data. The term "item of measurement information" as used herein is a broad term and should be given its usual and customary meaning for a person skilled in the art and should not be limited to a special or customized meaning. The term may refer in particular, without being limited, to knowledge or evidence providing a qualitative and / or quantitative description of at least one measurement, in particular at least one measurement object. The item of measurement information may include at least one of the physical properties of the measurement object and the chemical properties of the measurement object, in particular the chemical composition of the measurement object. The physical properties may in particular include optical properties such as at least one absorptivity of the measurement object and / or at least one emissivity of the measurement object. The chemical composition may in particular refer to qualitative and / or quantitative information about at least one material that the measurement object comprises. The evaluation device may be configured to perform a classification of the measurement object based on the item of measurement information. The item of measurement information, and thus the measurement object, may be assigned to one or more classes, for example classes predetermined by a user. Additionally or alternatively, the items of measurement information may be used to create and / or define at least one class, for example a new material class.

[0058] The transfer function may be updated, in particular recalibrated, by a user of the spectral measurement device or automatically by the spectral measurement device. For example, the evaluation device may be further configured to adapt the transfer function, such as when a calibration measurement is performed. The calibration measurement may be performed when changes in the operating and ambient conditions of the classification-based spectral measurement exceed predefined limits. In particular, the calibration measurement may be performed when at least one of the following occurs: the radiation output of the radiation source falls below a predefined threshold, for example due to degradation; the photodetector signal of the photodetector falls below a predefined threshold, for example due to degradation or hysteresis; the ambient temperature exceeds a predefined threshold and / or falls below a predefined threshold, the ambient humidity exceeds a predefined threshold and / or falls below a predefined threshold; the ambient light intensity exceeds a predefined threshold and / or falls below a predefined threshold. In particular, the evaluation device may be configured to update the transfer function accordingly if changes in the hardware and / or operating conditions are detected in the calibration measurement. As an example, the calibration measurement may reveal changes in the responsivity of at least one photodetector, for example due to degradation of at least one pixel. The weighting of the pixels can then be adapted in the transfer function, specifically by adapting the corresponding parameters of the transfer function.

[0059] For example, for automatic self-calibration, the spectral measurement device may be equipped as described in EP application no. 20211174.6 filed on 2 December 2020. The calibration measurement may be performed by using a sample interface. At least one of the photodetectors may be configured to generate at least one reference photodetector signal. In particular, at least one of the photodetectors may be configured to detect the optical radiation reflected from the sample interface and generate at least one corresponding reference photodetector signal. The sample interface may be configured to reflect the optical radiation from the radiation source without reaching the measurement object. In particular, the sample interface may be configured to direct the optical radiation emitted by the radiation source at least partially towards the photodetector such that the optical radiation does not pass through the measurement object. The sample interface material may be selected to reflect a portion of the light, for example 15%, 20% or 25%, depending on the classification application, and the reflected light from the sample interface is measured by the photodetector as a reference signal. For example, the sample interface may comprise an optical element, in particular an optical window, preferably selected from a glass window or a silicon window. For example, a beam splitter may be used. The optical element may be made of a transparent material, preferably a material that is at least partially transparent in at least one band of the wavelength range covered by the optical radiation. The optical element may be configured to guide a first portion of the optical radiation emitted by the radiation source directly to at least one of the optical detectors. A signal generated in response to this irradiation can then be used as a reference signal. Additionally or alternatively, instead of using Fresnel reflection of the window material, the spectral measurement device may comprise an integrated diffuse reflection target (also denoted integrated calibration target), which can be used as a calibration target when no sample is present. The integrated calibration target may be configured to guide the optical radiation emitted by the radiation source at least partially towards the optical detectors, in particular diffusely, in a predefined spectrum. The sample interface may in particular comprise precisely known and stable physical properties.

[0060] The spectrum measurement device may comprise at least one communication interface configured to transmit data from and / or to the evaluation device. In particular, the communication interface may be configured to receive instructions, commands, data, parameters, etc. for adapting the classification model and / or the transfer function. Thus, the classification model and / or the transfer function may be adapted by a user and / or by an external source with available updates, in particular with automatic updates. The term "communication interface" as used herein is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may in particular refer to an object or element forming a boundary configured to communicate information, without being limited thereto. In particular, the communication interface may be configured to communicate information from a computing device, e.g., a computer, e.g., to transmit or output information to another device. Additionally or alternatively, the communication interface may be configured to communicate information to a computing device, e.g., a computer, e.g., to receive information. The communication interface may in particular provide a means for communicating or exchanging information. In particular, the communication interface may provide a data transfer connection, e.g., Bluetooth, NFC, inductive coupling, etc. By way of example, the communication interface may be or may include at least one port, including one or more of a network or Internet port, a USB port, and a disk drive. The communication interface may further comprise at least one display device. The communication interface may be at least one web interface.

[0061] The spectral measurement device may comprise at least one temperature sensor. The temperature sensor may be configured to compensate the photodetector signal for ambient temperature. Additionally or alternatively, the spectral measurement device may comprise at least one temperature stabilization element, in particular a thermoelectric cooler. In particular, at least one of the photodetector and the radiation source may be arranged on and / or connected to the temperature stabilization element. The spectral measurement device may comprise at least one humidity sensor configured to compensate the photodetector signal for ambient humidity.

[0062] In a further aspect of the invention, there is provided a method for measuring optical radiation provided by at least one measurement object by performing a classification-based spectral measurement by using a spectral measurement device according to any one of the embodiments described above or disclosed in more detail below, the method comprising the following steps: emitting optical radiation at least partially towards the measurement object by using at least one radiation source, the optical radiation being at least partially in a spectral range that is subject to the classification-based spectral measurement; detecting optical radiation provided by the measurement object and generating at least one optical detector signal by using at least two optical detectors; measuring said photodetector signals and generating at least one item of spectral data by using at least one readout device; transferring said item of spectral data to at least one item of input data applicable to a classification model by using a transfer function; generating at least one item of measurement information by using said classification model; Includes.

[0063] The method comprises the steps of: performing at least one calibration measurement when changes in operating conditions and / or ambient conditions of the classification-based spectral measurement exceed predefined limits; fitting the transfer function corresponding to the calibration measurements; Further includes:

[0064] The method steps can be performed in a given order. However, it should be noted that different orders are possible. The method may include additional method steps not listed. Furthermore, one or more method steps may be performed once or repeatedly. Furthermore, two or more method steps may be performed simultaneously or with overlapping times. For further definitions and embodiments of the method, please refer to the definitions and embodiments of the spectrum measuring device.

[0065] The method may further include acquiring at least one classification model and at least one transfer function in the at least one evaluation device. The term "acquire" as used herein, including grammatical variations thereof, is a broad term and should be given its ordinary and customary meaning to those skilled in the art and should not be limited to a special or customized meaning. The term may specifically refer to selecting and / or loading an entity such as a model or function, without being limited thereto. Specifically, a plurality of classification models and / or transfer functions may be stored in a memory storage of the spectral measurement device. A user may then select at least one classification model and / or at least one transfer function to be used for the classification-based spectral measurement. The spectral measurement device may be configured to recommend at least one classification model based on the item of generated spectral data. Additionally or alternatively, the spectral measurement device may be configured to recommend at least one transfer function based on the hardware used, e.g., the photodetector, and / or the operating conditions, e.g., the temperature measured by a temperature sensor. Additionally or alternatively, the spectral measurement device may be configured to independently select at least one classification model based on the item of generated spectral data. Additionally or alternatively, the spectral measurement device may be configured to independently select the at least one transfer function based on the hardware used, e.g. the photodetector, and / or the operating conditions, e.g. the temperature measured by a temperature sensor. The at least one classification model and / or the at least one transfer function can be downloaded from at least one external source, e.g. by using a communication interface of the spectral measurement device.

[0066] The method can be computer-implemented. In particular, one or more of the method steps can be performed by using a computer or a computer network, more particularly by using a computer program. Thus, in general, any of the method steps involving providing and / or manipulating data can be performed by using a computer or a computer network. In general, these method steps can include any method steps, except those that typically require manual effort, such as providing a measurement object and / or certain aspects of performing the actual measurement.

[0067] Further disclosed and suggested herein is a computer program comprising instructions which, when the program is executed by a spectrum measurement device according to any one of the embodiments disclosed above or in more detail below with reference to the spectrum measurement device, cause the spectrum measurement device to perform a method according to any one of the embodiments disclosed above or in more detail below with reference to the method.

[0068] Further disclosed and proposed herein is a computer readable storage medium. The computer readable storage medium comprises instructions which, when the program is executed by a spectrum measurement device according to any one of the embodiments disclosed above or in more detail below referring to the spectrum measurement device, cause the spectrum measurement device to execute the method according to any one of the embodiments disclosed above or in more detail below referring to the method. The computer readable storage medium may refer to a non-transitory data storage means such as a hardware storage medium on which computer executable instructions are stored. The computer readable data carrier or storage medium may in particular be or include a storage medium such as a random access memory (RAM) and / or a read only memory (ROM).

[0069] In a further aspect of the invention, there is provided use of a spectral measurement device according to any one of the embodiments disclosed in further detail above or below with reference to the spectral measurement device for a purpose selected from the group consisting of: infrared detection applications; spectroscopy applications; exhaust gas monitoring applications; combustion process monitoring applications; pollution monitoring applications; industrial process monitoring applications; mixing or blending process monitoring applications; chemical process monitoring applications; food treatment process monitoring applications; food preparation process monitoring applications; water quality monitoring applications; air quality monitoring applications; quality control applications; temperature control applications; motion control applications; exhaust control applications; gas sensing applications; gas analysis applications; motion sensing applications; chemical sensing applications; mobile applications; medical applications; mobile spectroscopy applications; food analysis applications; agricultural applications, in particular soil, silage, feed, crop or agricultural product characterization, plant health monitoring; plastic identification and / or recycling applications; fiber identification and / or recycling applications.

[0070] The device and method according to the invention can offer a number of advantages over known methods, stations and systems. In particular, the device and method can allow both recording and analysis of spectral data by using a miniaturized device, in particular a handheld device. The results of the spectral measurement can therefore be presented to the user in a consumer-friendly and rapid manner. No extensive computing power is required to analyse the data and can be directly integrated in the handheld device. Furthermore, the data analysis can be easily adapted to different hardware or operating conditions by using a global classification model and a local transfer function that can take into account the hardware and operating conditions in question. The device and method are particularly low-cost, but also very reliable thanks to regular calibration of the hardware. Generally, such a calibration may require technical expertise, which is not consumer-friendly. However, here the calibration can be performed automatically, for example by using dark pixels of the photodetector. Additionally or alternatively, the calibration can be performed by using specified optical elements, in particular an internal calibration target. In this way, the user instructions and the required expertise can be minimized. Furthermore, potentially changing operating conditions can be automatically observed, for example by using the dark pixels or designated sensors, and taken into account for data analysis.

[0071] As used herein, the terms "having", "comprises", "including" or any grammatical variants thereof are used in a non-exclusive manner. Thus, these terms can refer both to the situation where no further features are present in the entity described in this context, in addition to the features introduced by these terms, and to the situation where one or more further features are present. As an example, the expressions "A has B", "A comprises B" and "A includes B" can refer both to the situation where no other elements are present in A besides B (i.e., the situation where A is composed exclusively of B), and to the situation where one or more further elements are present in entity A besides B, for example, the presence of element C, elements C and D, or further elements.

[0072] Furthermore, it should be noted that the terms "at least one," "one or more," or similar language indicating that a feature or element may be present more than once, are typically used only once when introducing each feature or element. In most cases, the language "at least one" or "one or more" will not be repeated when referring to each feature or element, regardless of the fact that each feature or element may be present one or more times.

[0073] Furthermore, as used herein, the terms "preferably", "more preferably", "particularly", "more particularly", "particularly", "even more particularly" or similar terms are used in combination with any feature without limiting the possibilities of substitution. Thus, features introduced by these terms are optional features and are not intended to limit the scope of the claims in any way. The invention may also be practiced by using alternative features, as will be recognized by those skilled in the art. Similarly, features introduced by "in one embodiment of the invention" or similar expressions are intended to be optional features, without limitations on alternative embodiments of the invention, without limitations on the scope of the invention, and without limitations on the possibilities of combining the feature introduced in such a manner with other optional or non-optional features of the invention.

[0074] In summary, without excluding further possible embodiments, the following embodiments can be envisaged:

[0075] Embodiment 1. A spectral measurement device for measuring optical radiation provided by at least one measurement object with a spectral measurement based on at least one classification, the spectral measurement device comprising: - at least one radiation source configured to emit optical radiation at least partially towards the measurement object, the optical radiation being at least partially within a spectral range that is subject to a classification-based spectral measurement; - at least two photodetectors, each photodetector comprising at least one pixel, each pixel being an active pixel or a dark pixel, said spectral measurement device comprising at least two active pixels, said spectral measurement device comprising at least one dark pixel, each active pixel configured to generate at least one photodetector signal dependent on illumination of said pixel, at least two of said active pixels configured to detect optical radiation in at least partially different spectral ranges, each dark pixel configured to generate at least one photodetector signal independent of illumination of said dark pixel; - at least one readout device configured to measure said photodetector signal and generate at least one item of spectral data; at least one evaluation device comprising at least one processor and at least one memory storage, said memory storage comprising: at least one classification model including a set of distinct classes, each class referring to at least one spectral characteristic, said classification model configured to classify at least one item of input data into said class; at least one transfer function configured to transfer the items of spectral data to items of input data applicable to the classification model; At least one evaluation device configured to store Equipped with A spectral measurement device, wherein the evaluation device is configured to generate at least one item of measurement information by applying the classification model and the transfer function to an item of the spectral data.

[0076] Embodiment 2. The spectral measurement device of the preceding embodiment, wherein each photodetector includes at most one active pixel, the spectral measurement device includes at least one dark pixel, and the evaluation device is configured to automatically recalibrate the spectral measurement device by using at least one photodetector signal generated by using the dark pixel.

[0077] Embodiment 3. A spectral measurement device as described in any one of the preceding embodiments, wherein each active pixel is configured to detect optical radiation having at least one spectral characteristic that refers to at least one class of the classification model.

[0078] Embodiment 4. A spectral measurement device according to any one of the preceding embodiments, wherein each class refers to at least one material having at least one distinct spectral characteristic, or at least one group of materials having similar spectral characteristics.

[0079] Embodiment 5. The spectral measurement device of the preceding embodiment, wherein the material is at least one material selected from the group consisting of synthetic fibers; cotton; wool; silk; polyethylene terephthalate (PET); polypropylene (PP); polyethylene (PE); polyvinyl chloride (PVC); high density polyethylene (HDPE); low density polyethylene (LDPE); polyamide (PA); and glass.

[0080] Embodiment 6. A spectral measurement apparatus according to any one of the preceding embodiments, wherein each spectral characteristic refers to at least one predetermined and / or predefined spectral range.

[0081] Embodiment 7. The spectral measurement device of the preceding embodiment, wherein the predetermined and / or predefined spectral range is at least one spectral range selected from the group consisting of: 1650 nm to 1700 nm; 1725 nm to 1800 nm; 2050 nm to 2150 nm; and 2175 nm to 2225 nm.

[0082] Embodiment 8. A spectral measurement device as described in any one of the preceding embodiments, wherein the classification model is based on at least one algorithm trained on past spectral data acquired by at least one external device, in particular at least one further spectral measurement device.

[0083] Embodiment 9. A spectrum measurement device as described in any one of the preceding embodiments, wherein the classification model is generated by using at least one of a random forest, a K-means clustering, and an SVM.

[0084] Embodiment 10. A spectral measurement device as described in any one of the preceding embodiments, wherein the transfer function includes at least one parameterized mathematical function including at least one parameter, and the transfer function takes into account at least one of the hardware of the spectral measurement device and the operating conditions of the spectral measurement device.

[0085] Embodiment 11. A spectral measurement device as described in the preceding embodiment, wherein parameters of the transfer function are selected to weight the photodetector signal.

[0086] Embodiment 12. A spectral measurement device according to any one of the preceding two embodiments, wherein the parameterized mathematical function includes at least one matrix.

[0087] Embodiment 13. A spectrum measurement device according to any one of the preceding embodiments, wherein the transfer function includes at least one lookup table.

[0088] Embodiment 14. A spectral measuring device described in any one of the preceding embodiments, wherein the items of measurement information include at least one of a physical property of the measurement object and a chemical property of the measurement object, specifically a chemical composition of the measurement object.

[0089] Embodiment 15. A spectral measurement device as described in any one of the preceding embodiments, wherein the evaluation device is configured to perform a classification of the measurement object based on the items of measurement information.

[0090] Embodiment 16. A spectral measurement device according to any one of the preceding embodiments, wherein the spectral measurement device is configured to perform at least one calibration measurement.

[0091] Embodiment 17. The spectral measurement device of the preceding embodiment, wherein the evaluation device is further configured to adapt the transfer function when a calibration measurement is performed.

[0092] Embodiment 18. A spectral measurement device as described in any one of the preceding embodiments, wherein the radiation source comprises at least one of a semiconductor-based radiation source and a thermal emitter.

[0093] Embodiment 19. The spectral measurement device of the preceding embodiment, wherein the semiconductor-based radiation source is selected from at least one of a light emitting diode (LED) or a laser, specifically a laser diode.

[0094] Embodiment 20. A spectral measurement device according to any one of the preceding embodiments, wherein the radiation source is modulated.

[0095] Embodiment 21. A spectral measurement device as described in any one of the preceding embodiments, wherein at least one of the photodetectors is a PbS photodetector and the emitted optical radiation includes wavelengths of 400 nm to 3000 nm, specifically 800 nm to 2700 nm, more specifically 1200 nm to 2500 nm.

[0096] Embodiment 22. A spectral measurement device as described in any one of the preceding embodiments, wherein at least one of the photodetectors is a PbSe photodetector and the wavelength of the emitted optical radiation includes wavelengths of 400 nm to 5000 nm, specifically 800 nm to 4800 nm, and more specifically 1200 nm to 4600 nm.

[0097] Embodiment 23. A spectral measurement device as described in any one of the preceding embodiments, wherein the spectral measurement device comprises at least two radiation sources configured to emit optical radiation in at least partially different spectral ranges.

[0098] Embodiment 24. A spectral measurement device as described in the preceding embodiment, wherein each optical detector is different and configured to detect optical radiation emitted by exactly one of the radiation sources.

[0099] Embodiment 25. A spectral measurement device as described in any one of the preceding embodiments, wherein the spectral measurement device comprises at least three photodetectors, specifically at least four photodetectors, more specifically at least five photodetectors.

[0100] Embodiment 26. A spectral measurement device as described in any one of the preceding embodiments, wherein the spectral measurement device comprises at least three pixels.

[0101] Embodiment 27. A spectral measurement device according to any one of the preceding embodiments, wherein the spectral measurement device comprises at least two active pixels and at least one dark pixel.

[0102] Embodiment 28. A spectral measurement device as described in any one of the preceding embodiments, wherein at least one of the photodetectors comprises at least two pixels.

[0103] Embodiment 29. A spectral measurement device as described in the preceding embodiment, wherein at least one of the pixels is an active pixel and at least one of the pixels is a dark pixel.

[0104] Embodiment 30. A spectral measurement device according to any one of the preceding two embodiments, wherein the pixels comprise the same photosensitive material.

[0105] Embodiment 31. A spectral measurement device as described in any one of the preceding embodiments, wherein at least two of the pixels are different sizes.

[0106] Embodiment 32. A spectral measurement device according to any one of the preceding embodiments, wherein the size of each pixel is such that at least similar photodetector signal intensity is produced by each photodetector.

[0107] Embodiment 33. A spectral measurement device described in any one of the preceding embodiments, wherein at least one optical detector is equipped with at least one optical filter, specifically an optical bandpass filter, more specifically a narrow bandpass (NBP) optical filter.

[0108] Embodiment 34. A spectral measurement device described in any one of the preceding embodiments, wherein at least one of the radiation source and / or the optical detector is positioned such that directly reflected optical radiation from the measurement object does not reach at least one of the active pixels.

[0109] Embodiment 35. A spectral measurement device described in any one of the preceding embodiments, wherein at least one of the optical detectors is configured to measure diffusely reflected optical radiation from the measurement object.

[0110] Embodiment 36. A spectral measurement device as described in any one of the preceding embodiments, wherein at least one of the photodetectors is insulated by at least one insulating material.

[0111] Embodiment 37. A spectral measurement device as described in any one of the preceding embodiments, wherein at least one of the photodetectors, specifically at least one of the pixels, includes at least one photoconductive material.

[0112] Embodiment 38. A spectral measurement device as described in the preceding embodiment, wherein the photoconductive material is selected from at least one of PbS, PbSe, Ge, InGaAs, InSb, or HgCdTe.

[0113] Embodiment 39. A spectral measurement device as described in any one of the preceding embodiments, wherein the measurement object includes at least one solid sample.

[0114] Embodiment 40. A spectral measurement device as described in any one of the preceding embodiments, wherein the spectral measurement device comprises at least one sample interface configured to at least partially direct optical radiation emitted by the radiation source toward the optical detector in a predefined spectrum.

[0115] Embodiment 41. A spectral measurement device as described in the preceding embodiment, wherein at least one of the optical detectors is configured to detect optical radiation reflected from the sample interface and generate at least one corresponding reference optical detector signal.

[0116] Embodiment 42. A spectral measurement device as described in any one of the preceding embodiments, wherein the readout device comprises at least one of a resistance meter, a voltmeter, an ammeter, a lock-in amplifier, a voltage divider, and an electrical pass filter.

[0117] Embodiment 43. A spectral measurement device as described in any one of the preceding embodiments, wherein the spectral measurement device includes at least one temperature sensor configured to compensate the photodetector signal for ambient temperature.

[0118] Embodiment 44. A spectral measurement device according to any one of the preceding embodiments, wherein the spectral measurement device comprises at least one temperature stabilizing element, in particular a thermoelectric cooler.

[0119] Embodiment 45. A spectral measurement device as described in any one of the preceding embodiments, wherein the spectral measurement device comprises at least one humidity sensor configured to compensate the photodetector signal for ambient humidity.

[0120] Embodiment 46. A spectral measurement device as described in any one of the preceding embodiments, wherein the spectral measurement device comprises at least one communication interface configured to transmit data from and / or to the evaluation device, in particular to adapt the classification model and / or the transfer function, and the spectral measurement device comprises at least one optical element, in particular at least one integrated calibration target, configured to at least partially direct the optical radiation emitted by the radiation source towards the photodetector in a predefined spectrum.

[0121] Embodiment 47. A spectral measurement device as described in the preceding embodiment, wherein the optical element is configured to at least partially reflect the optical radiation emitted by the radiation source, in a predefined spectrum, particularly diffusely, towards the optical detector.

[0122] Embodiment 48. A method for measuring optical radiation provided by at least one measurement object by performing classification-based spectral measurements by using a spectral measurement device according to any one of the preceding embodiments, the method comprising: emitting optical radiation at least partially towards the measurement object by using at least one radiation source, the optical radiation being at least partially in a spectral range that is subject to a spectral measurement based on the classification; detecting the optical radiation provided by the measurement object and generating at least one optical detector signal by using at least two optical detectors; measuring said photodetector signals and generating at least one item of spectral data by using at least one readout device; transferring said item of spectral data to at least one item of input data applicable to a classification model by using a transfer function; generating at least one item of measurement information by using said classification model; A method comprising:

[0123] 49. performing at least one calibration measurement when a change in operating and / or ambient conditions of the classification-based spectral measurement exceeds a predefined limit; fitting the transfer function corresponding to the calibration measurements; 4. The method of claim 1, further comprising:

[0124] Embodiment 50. A method according to any one of the preceding method embodiments referring to a method, wherein the method includes obtaining at least one classification model and at least one transfer function in at least one evaluation device.

[0125] Embodiment 51. A method according to any one of the preceding method embodiments, wherein the method is computer-implemented.

[0126] Embodiment 52. A computer program comprising instructions, when the program is executed by a spectrum measuring device described in any one of the preceding embodiments referring to a spectrum measuring device, causing the spectrum measuring device to perform a method described in any one of the preceding embodiments referring to a method.

[0127] Embodiment 53. A computer-readable storage medium comprising instructions, which when executed by a spectrum measurement device described in any one of the preceding embodiments referring to a spectrum measurement device, causes the spectrum measurement device to perform the method described in any one of the preceding embodiments referring to the method.

[0128] Embodiment 54. Use of a spectral measuring device according to any one of the preceding embodiments referring to a spectral measuring device, the use being for a purpose selected from the group consisting of: infrared detection applications; spectroscopy applications; exhaust gas monitoring applications; combustion process monitoring applications; pollution monitoring applications; industrial process monitoring applications; mixing or blending process monitoring applications; chemical process monitoring applications; food treatment process monitoring applications; food preparation process monitoring applications; water quality monitoring applications; air quality monitoring applications; quality control applications; temperature control applications; motion control applications; exhaust control applications; gas detection applications; gas analysis applications; motion sensing applications; chemical sensing applications; mobile applications; medical applications; mobile spectroscopy applications; food analysis applications; agricultural applications, in particular soil, silage, feed, crop or agricultural product characterization, plant health monitoring; plastic identification and / or recycling applications; fiber identification and / or recycling applications. [Brief description of the drawings]

[0129] Further optional features and embodiments are disclosed in more detail in the description following the embodiments, preferably in conjunction with the dependent claims, where each optional feature may be realized in an isolated manner as well as in any possible combination as understood by a person skilled in the art. The scope of the present invention is not limited by the preferred embodiments. The embodiments are illustrated diagrammatically in the figures, where identical reference numbers in these figures refer to identical or functionally equivalent elements.

[0130] In the diagram: [Figure 1] 1 shows a schematic diagram of an exemplary embodiment of a spectral sensing device; [Diagram 2] 2A and 2B show an arrangement of multiple single pixels with a single radiation source and with wavelength dependent optical filters. [Diagram 3] 3A and 3B show an embodiment in which all pixels are twins. [Figure 4] Material specific absorption bands for different fiber classes are shown. [Diagram 5] 5A and 5B show an assembly with a centrally located radiation source and an assembly with multiple peripherally located radiation sources. [Figure 6] 1 shows a schematic diagram of an exemplary embodiment of an optical radiation measuring method according to the invention; DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0131] Detailed Description of the Preferred Embodiments FIG. 1 shows in a highly schematic manner an exemplary embodiment of a spectrum measurement device 110 according to the present invention. The spectrum measurement device 110 is a device configured to determine spectral information by recording at least one measurement of at least one signal intensity associated with at least one corresponding signal wavelength of the optical radiation 112 and by evaluating at least one photodetector signal associated with the signal intensity. In the embodiment shown in FIG. 1, the spectrum measurement device 110 comprises a housing 114 that encloses the components of the spectrum measurement device 110. In this manner, the components of the spectrum measurement device 110 can be protected and access to external light can be prevented. However, further types of arrangements of the components of the spectrum measurement device 110 can be envisaged.

[0132] As mentioned above, the spectral measurement device 110 is configured to measure optical radiation with a spectral measurement based on at least one classification. A class may be a group of entities, e.g., objects or elements, sharing at least one characteristic. The shared characteristic may refer to a physical characteristic, specifically an optical characteristic, such as a spectral characteristic, of the object or element. The shared characteristic may further refer to a chemical characteristic, such as a chemical composition of the object. As an example, the shared characteristic of the objects may be a common material that they all contain. Specifically, each class may refer to at least one material having at least one distinct spectral characteristic, or at least one group of materials having similar spectral characteristics. The spectral characteristic may specifically include at least one absorption peak of the material and / or a physical quantity related to the absorption peak. More specifically, the material may be at least one material selected from the group consisting of synthetic fibers; cotton; wool; silk; polyethylene terephthalate (PET); polypropylene (PP); polyethylene (PE); polyvinyl chloride (PVC); high density polyethylene (HDPE); low density polyethylene (LDPE); polyamide (PA); glass. In general, the term category may be used synonymously with the term class. Classification may include first analyzing the entities and identifying shared properties. As an example, classification may specifically refer to analyzing the chemical composition of the objects and grouping objects containing a common material into one class. Classification may refer to the creation and / or definition of classes, specifically defining class boundaries. As an example, classification may set at least one limit of a physical property, such as at least one of density, enrichment, concentration, as a class boundary. As a further example, classification may set at least one maximum deviation from a predefined condition or state as a class boundary. For example, a measured object that is mainly composed of one material except for a predefined deviation portion can be grouped into one class. In this way, the classes can be specifically defined by material. Classification may include assigning the entities to at least one class. In principle, it is also possible to assign an entity to multiple classes simultaneously.Thus, given a defined class, at least one entity can be assigned to the defined class, in particular if that entity shares with other entities at least one characteristic that the class already contains, or if that entity is within the boundaries of the defined class. As an example, the classes can be defined by material. The measured object can then be assigned to a class or category if it contains material at least to a predetermined extent (which can be determined, for example, by measuring at least one physical property of the object, such as an optical property, e.g. absorption). The classification-based spectral measurement can be a spectral measurement, where the measured spectral data and / or the preprocessed measured spectral data are subsequently classified. The result or intermediate result of the classification-based spectral measurement can in particular include information regarding the class to which the measured object is assigned.

[0133] The spectral measurement device 110 comprises at least one radiation source 116 configured to emit optical radiation 112 at least partially towards a measurement object 118, said optical radiation 112 being at least partially within a spectral range that is subject to a classification-based spectral measurement. In particular, the radiation source 116 may be preferably comprised by a semiconductor-based radiation source 116 that may be selected from at least one of a light-emitting diode (LED) or a laser, in particular a laser diode. However, further types of radiation source 116 may also be possible. The radiation source 116 may emit or generate modulated light pulses continuously.

[0134] The exemplary spectrum measurement device 110 includes at least two photodetectors 120 .

[0135] The photodetectors 120 may comprise at least one substrate. A single photodetector 120 may be a substrate having at least one single photosensitive area that generates a physical response to illumination in a given wavelength range. Each photodetector 120 comprises at least one pixel 121. An embodiment of a pixel is shown in Fig. 2 and Fig. 3. Each photodetector 120 is configured to generate at least one photodetector signal depending on illumination of the pixel 121. At least two of the photodetectors 120 are configured to detect optical radiation in at least partially different spectral ranges. The pixels 121 may be illuminated by optical radiation and may generate at least one photodetector signal in response to illumination. The pixels 121 may be arranged on a surface of the photodetector 120, in particular on a substrate. Furthermore, the pixels 121 may be in particular a single, closed, uniform photosensitive area.

[0136] At least two of the pixels 121 may be different in size. The sizes of the pixels may differ from each other. The pixel sizes from different substrates or even the pixel sizes of the dual pixel detector may be different. The size of each pixel 121 may be such that at least similar photodetector signal intensities are generated by each photodetector 120, taking into account the photosensitivity of each pixel at a given wavelength and the intensity of the optical radiation incident on each pixel. In this way, it may be possible to obtain similar signal intensities from multiple detectors during a measurement, which may help in optimizing the readout electronics. Other embodiments of the pixels 121 may also be possible.

[0137] The pixels 121 of each photodetector 120 can be manufactured from the same material. Alternatively, pixels of different pixel materials can be combined into a single detector assembly. In this way, the spectral measurement device can cover a wide wavelength range and can be optimized in terms of performance, for example by varying the size of the active area. As an example, the single detector assemblies can have PbS and PbSe pixels of different sizes on different substrates.

[0138] The pixels 121 may preferably comprise at least one inorganic photoconductive material selected from photoconductive materials, in particular lead sulfide (PbS), lead selenide (PbSe), germanium (Ge), indium gallium arsenide (InGaAs, including but not limited to extended InGaAs), indium antimony (InSb), or mercury cadmium telluride (HgCdTe or MCT). In particular, at least one of the photodetectors 120 may be a PbS photodetector, and the emitted optical radiation 112 comprises wavelengths between 760 400 nm and 3000 nm, in particular between 1000 800 nm and 2700 nm, more particularly between 1500 1200 nm and 2500 nm. At least one of the photodetectors 120 may be a PbSe photodetector, with the wavelength of the emitted optical radiation 120 including wavelengths between 400 nm and 5000 nm, specifically between 800 nm and 4800 nm, more specifically between 1200 nm and 4600 nm, however, different types of materials or other types of photodetectors 120 may also be possible.

[0139] The spectral measurement device 110 may comprise at least one readout device 139 configured to measure the photodetector signals and generate at least one item of spectral data. The readout device 139 may comprise at least one of an ohmmeter, a voltmeter, an ammeter, a lock-in amplifier, a voltage divider, and an electrical pass filter. The item of spectral data may relate to the absorption of optical radiation by the measurement object 118. The item of spectral data may include at least one absorption spectrum of the measurement object 118, which may include at least one absorption peak. The item of spectral data may be analog and / or digital. The readout device 139 may be configured to collect detector signals, each detector signal may be indicative of, for example, a detected intensity of optical radiation in a different spectral range. The readout device 139 may be configured to process and / or pre-process the detector signals. The readout device may be configured to combine the detector signals in at least one spectrum, in particular at least one absorption spectrum of the measurement object 118.

[0140] At least two of the photodetectors 120 are configured to detect optical radiation in at least partially different spectral ranges. The spectral ranges are not identical, and the spectral ranges may or may not overlap. At least two of the photodetectors 120 may be different, e.g., may include pixels including different photoconductive materials. In particular, each active pixel 123 may be configured to detect optical radiation having at least one spectral characteristic that refers to at least one class of the classification model. Each active pixel 123 may be configured to detect optical radiation having at least one spectral characteristic that refers to exactly one class of the classification model. Each class may in particular refer to at least one material having at least one distinct spectral characteristic, or at least one group of materials having similar spectral characteristics. Each spectral characteristic may refer to at least one predetermined and / or predefined spectral range. The spectral characteristic, e.g. an absorption peak of a material, may be located within the predetermined and / or predefined spectral range. The predetermined and / or predefined spectral range may be pre-determined and / or pre-defined before measuring the optical radiation provided by the measurement object, in particular according to a pre-determined classification. As an example, a classification-based spectral measurement may be envisaged to determine the material in the garment. Before measuring the optical radiation provided by the garment, a fiber class may be pre-determined and / or pre-defined, each fiber class may include a spectral characteristic such as an absorption peak in a different predetermined and / or predefined spectral range. The predetermined and / or predefined spectral range may be at least one spectral range selected from the group consisting of: 1650 nm to 1700 nm; 1725 nm to 1800 nm; 2050 nm to 2150 nm; 2175 nm to 2225 nm. However, other predetermined and / or predefined spectral ranges are possible in other use cases.

[0141] Each pixel is an active pixel 123 or a dark pixel 124. The substrate may have only one light-sensitive area or a dual light-sensitive area, with one "active" area and one second "dark" area, by mechanical or optical means. The spectral measurement device 110 comprises at least two active pixels 123. Each active pixel 123 is configured to generate at least one photodetector signal that depends on the illumination of the active pixel 123. At least two of the active pixels 123 are configured to detect optical radiation in at least partially different spectral ranges. An active pixel 123 may be a pixel that is arranged and / or configured to generate a photodetector signal in response to illumination of its light-sensitive area. Active may refer to the light-sensitive area being illuminated in a sensitivity wavelength range during the measurement. The spectral measurement device 110 comprises at least one dark pixel 124. Each dark pixel 124 is configured to generate at least one photodetector signal that does not depend on the illumination of the dark pixel 124. The dark pixel 124 may be a covered pixel 121. Dark may refer to the fact that no radiation in the wavelength range of sensitivity of the photosensitive area reaches the photosensitive area. The dark pixels 124 may be covered by a material that absorbs in the same wavelength range as the wavelength range of sensitivity of the photodetector and / or may be metal pieces on the photosensitive area that reflect the incident light from the dark pixels. The dark pixels 124 may be covered with a material that absorbs the optical radiation, such as glue and / or ink. Additionally or alternatively, the dark pixels may be covered with at least one material that reflects the optical radiation, such as metal. The dark pixels 124 may in particular not see any optical radiation at all, or at least hardly see any optical radiation. The dark pixels 124 may be embodied as an additional photodetector with a single pixel covered by the above mentioned means, or a photodetector with dual pixels 121 (on one single substrate) may be used, where one of the pixels 121 is darkened. The dark pixels 124 may be used to eliminate long-term drift and temperature effects.

[0142] Each photodetector 120 may include at most one active pixel 123, and the spectral measurement device 110 may include at least one dark pixel 124. For example, each of the photodetectors 120 may include two pixels 121, namely an active pixel 123 and a dark pixel 124. For example, one of the photodetectors 120 may include a single active pixel 123 and another photodetector 120 may include a dark pixel 124. None of the photodetectors 120 may have more than one active pixel 123. Each active pixel 123 may be configured to detect a specific spectral range and / or wavelength range. Thus, each photodetector 120 may be responsible for only a specific spectral range.

[0143] At least one of the photodetectors 120 may comprise at least two pixels 121. At least one of the pixels 121 may be an active pixel 123. At least one of the pixels 121 may be a dark pixel 124. Specifically, the photodetector 120 may be a dual pixel photodetector. Thus, the photodetector 120 may comprise an active pixel 123 and a dark pixel 124. The pixels 121 may comprise the same photosensitive material. Thus, the calibration performed by using the dark pixel 124 may be based on the same photosensitive material that the active pixel 123 comprises. All the pixels 121 on one photodetector 120 may comprise the same photosensitive material.

[0144] The spectral measurement device 110 may comprise at least two radiation sources 116 configured to emit light radiation in at least partially different spectral ranges. As an example, the spectral measurement device 110 may comprise two different radiation sources 116, for example two different LEDs. Additionally or alternatively, the two radiation sources 116 may comprise different light filters.

[0145] Each photodetector 120 may be different and configured to detect optical radiation emitted by exactly one radiation source 116. Each photodetector 120 may be assigned to observe a different spectral range. Each photodetector 120 may be assigned to observe a spectral characteristic corresponding to a different class of the classification model. The spectrum measurement device 110 may comprise at least three photodetectors 120, specifically at least four photodetectors 120, more specifically at least five photodetectors 120. For example, the spectrum measurement device 110 may comprise 16 or fewer photodetectors 120. Specifically, the spectrum measurement device 110 may comprise at least three pixels 121. More specifically, the spectrum measurement device 110 may comprise at least two active pixels 123 and at least one dark pixel 124. Each active pixel 123 may be comprised by a different photodetector 120. Each of the two active pixels 123 may be used to observe a different spectral range. The dark pixels 124 may be used to calibrate the spectrum measurement device 110 .

[0146] At least one of the photodetectors 120 may be configured to generate at least one reference photodetector signal. In particular, at least one of the photodetectors 120 may be configured to detect the optical radiation reflected from the sample interface and generate at least one corresponding reference photodetector signal. The exemplary spectrum measurement device 110 may comprise optical elements, in particular an optical window 126, preferably selected from a glass window or a silicon window. A beam splitter may be used. The transparent material contained by the optical window 126 may preferably be at least partially transparent in at least one band of the wavelength range covered by the optical radiation 112. As shown diagrammatically, the optical elements may be configured to direct a first portion 128 of the optical radiation 112 emitted by the radiation source 116 to at least one of the photodetectors 120. The optical elements may further be configured to direct a second portion 130 of the optical radiation 112 emitted by the radiation source 116 to the measurement object 118. The measurement object 118 may be or include any object selected from living and non-living objects, including materials for investigation or monitoring by the spectral measurement device 110. As a result of this particular assembly of the radiation source 116, the optical elements, and the photodetectors 120 in the spectral measurement device 110, the optical elements may function as beam splitting elements designated to split the optical radiation 112 as emitted by the radiation source 116 into a first portion 128 that is directed directly towards the photodetectors 120 and a second portion 130 that is directed indirectly via the measurement object 118 towards the at least one photodetector 120.

[0147] The optical element may reflect a first portion 128 of the optical radiation 112 to the photodetector 120 and transmit a second portion 130 to the at least one measurement object 118. However, in an alternative embodiment (not shown here), the optical element may transmit the first portion 128 of the optical radiation 112 to the photodetector 120 and reflect the second portion 130 to the measurement object 118. Regardless of the embodiment selected, the optical radiation traverses an optically transparent medium 134 on the path from the emitting element 116 to the optical element. Here, the optically transparent medium 134 may be selected from, inter alia, ambient air, an inert gas or a vacuum, although further materials may also be possible as indicated in more detail above.

[0148] The spectral measurement device 110 may further include a spectral transfer element 136 configured to select at least one wavelength of the optical radiation 112 of either the first portion 128 and / or the second portion 130 for illuminating the optical detector 120. As shown diagrammatically here, the spectral transfer element 136 may be an optical filter 138, such as, inter alia, a polarizing filter; or a bandpass filter, although further types of spectral transfer element 136 are possible. In further embodiments (not shown here), the spectral transfer element 136 is not necessary.

[0149] At least one of the photodetectors 120 may include at least one optical filter 138. For example, the optical filter 138 may be an optical bandpass filter. For example, the optical filter 138 may be a narrow bandpass (NPB) optical filter. The spectral range detected by each photodetector 120 may be limited to a spectral range associated with at least one class of the classification, e.g., exactly one class of the classification. The photodetectors 120 may be assigned to observe optical characteristics associated with at least one class of the classification, specifically, exactly one class of the classification. Each photodetector 120 of the spectrum measurement device 110 may be assigned to observe a different spectral range. Each photodetector 120 may be assigned to observe spectral characteristics corresponding to a different class of the classification model. The optical filter 138 may be any device or object configured to selectively block optical radiation according to at least one physical characteristic of the optical radiation, such as the wavelength of the optical radiation and / or the polarization of the optical radiation. The optical bandpass filter may be an optical filter configured to filter incident optical radiation such that only a limited spectral range can pass through the optical filter and propagate to the object behind the optical filter. The optical bandpass filter may be configured to transmit only optical radiation from a minimum wavelength to a maximum wavelength. The spectral measurement device 110 may be designed specifically for each use case. The spectral measurement device 110 may be configured to measure application-specific absorption bands. The optical detector 120 may be configured to the absorption bands and may include an optical filter 138, such as a narrow bandpass (NBP) optical filter, with application-specific center wavelength, full width at half maximum (FWHM), out-of-area cutoff level and range. Additionally or alternatively, the optical detector 120 may be configured to those application-specific absorption bands by selecting an appropriate radiation source, such as an LED, that emits sample-specific wavelengths.

[0150] As further illustrated diagrammatically in Fig. 1, the spectrum measurement device 110 further comprises an evaluation device 140. The evaluation device 140 comprises at least one processor 142 and at least one memory storage 144. The memory storage 144 is configured to store at least one classification model comprising a set of different classes, each class referring to at least one spectral characteristic, the classification model being configured to classify at least one item of input data into the class. The memory storage 144 is configured to store at least one transfer function configured to transfer an item of spectral data to an item of input data applicable to the classification model. The evaluation device 140 is configured to generate at least one item of measurement information by applying the classification model and the transfer function to the item of spectral data.

[0151] The classification model is configured to classify at least one item of the input data into a class. The classification model may be a model or scheme that can be used to perform at least one classification. The classification model may include pre-determined and / or pre-defined classes, such as classes pre-determined and / or pre-defined by a user. The classification model may include class boundaries. The classification model may include class properties, such as at least one property shared by all entities in a class. The classification model may classify items of the input data into exactly one class. The classification model may classify items of the input data into multiple classes.

[0152] The classification model may be a global classification model. The global classification model may be a hardware-independent but application-specific classification model. The classification model may be use-case specific wavelength-, especially spectral range-, dependent. For every application there may be a unique global model. The classification model may be hardware and / or operating condition independent. Thus, the classification model may be globally applicable, especially for different hardware components, e.g. photodetectors used under different operating conditions such as temperature.

[0153] The classification model may include at least one algorithm for analyzing items of input data and classifying items of input data into classes. The classification model may include at least one trained model. The trained model is trained on at least one training set, in particular by using machine learning. The training data set may include, for example, past spectral data acquired by at least one external device, in particular at least one further spectral measurement device. The further spectral measurement device may be a laboratory spectral measurement device. The classification model may be generated by using at least one of Random Forest, K-Means Clustering, and Support Vector Machine (SVM), in particular depending on the application and the subsequent accuracy.

[0154] The classification model is configured to classify at least one item of input data into a class. The item of input data may be at least one of knowledge or evidence providing a qualitative and / or quantitative description relating to the information, recorded and / or structured in such a way that the information is interpreted, in particular for further processing, in at least one entity, e.g., in a model, program, processor, or network. In particular, the item of input data is applicable to the classification model as described above. The classification model may be configured to interpret the item of input data, to process the item of input data, in particular to classify the item of input data and / or the item of processed input data. The classification model may be particularly application-oriented. As an example, the classification may refer to a material analysis of clothing, as described above, and the classes may be fiber classes, such as cotton or wool. In this example, the classification model may be configured to classify input data in the form of processed spectral data referring to clothing into fiber classes.

[0155] The memory storage 144 is configured to store at least one transfer function. The transfer function is configured to transfer an item of spectral data to an item of input data applicable to the classification model. The global classification model can be adopted to similar hardware designed for its specific use case with the help of the transfer function. The data measured by the spectrum measurement device strongly depends on the tolerances of the hardware as well as on the ambient conditions such as temperature and humidity. As an example, the readout device 139 of the spectrum measurement device 110 can generate an item of spectral data by a spectral measurement of the measurement object 118. The item of spectral data may first depend on the measurement object 118, for example the absorption characteristics of the measurement object 118. However, the item of spectral data may further depend on the spectrum measurement device 110 itself, specifically the hardware of the spectrum measurement device 110, such as the photodetector used. Furthermore, the item of spectral data may depend on the operating conditions (such as temperature) of the spectrum measurement device 110 during the spectrum measurement. The transfer function transforms the data measured by the spectrum measurement device 110 such that the data can be applied to the global classification model offline on a similar device. Similar devices may refer to predefined hardware having predefined filters, number of pixels, pixel shapes, and the like.

[0156] The transfer function may be any function or algorithm for transferring data. The transfer may include at least one transformation of the data, in particular adapting the data for further processing, more particularly adapting the data for further processing within at least one particular entity, such as a particular model. The transfer may include transferring the data in such a way that information contained by the data is at least partially maintained. In particular, information relevant for further evaluation of the data is maintained, but the format of the data may be changed. Additionally or alternatively, further information may be added. For example, information about the hardware and / or operating conditions may be added to the item of spectral data to take into account the hardware and / or operating conditions for further evaluation, and / or the item of spectral data may be adapted by using the transfer function, in particular by taking into account the information about the hardware and / or operating conditions when generating the at least one item of input data applicable to the classification model. The transfer function may include at least one parameterized mathematical function including at least one parameter. The transfer function may be a function of coefficients corresponding to the pixels. The transfer function may take into account at least one of the hardware of the spectrum measurement device 110 and the operating conditions of the spectrum measurement device 110. In particular, the parameters of the transfer function may be selected to weight the photodetector signals, more particularly depending on at least one of the applied shape of the pixel and the applied material of the pixel. As an example, the spectrometer device 110 may comprise at least two pixels 121 having different shapes and / or different active materials. The two pixels 121 may thus induce different photodetector signals for the same incident light radiation containing the spectral information of one measurement object. By using a transfer function to transfer an item of spectral data to an item of input data applicable to the classification model, the different shapes and / or different active materials may be taken into account when classifying the measurement object 118, for example, into at least one material class by using the classification model.The transfer function allows the material classification application to be localized to the hardware, enabling mass deployment. The transfer function can be derived from an item of known spectral data, for example obtained by using a reference device on a reference sample. The reference device can comprise, for example, a high-end lab device. The reference sample can comprise, for example, a known sample, such as a known material. In particular, the transfer function can include at least one pre-determined coefficient. The parameterized mathematical function can include at least one matrix M of: Y=MX, where Y is a vector containing items of input data and X is a vector containing items of spectral data. The predetermined coefficients may thus be, for example, elements of a matrix M. The parameters of the matrix M may be determined by comparing the items of measured spectral data with items of known spectral data, e.g. obtained by using a reference device on a reference sample. The matrix M may, for example, contain device characteristics, e.g. detection rate, or temperature or humidity compensation parameters. The transfer function may further comprise at least one look-up table. The look-up table may be configured to match items of spectral data with items of input data applicable to the classification model.

[0157] As a particular example, the transfer function may allow for temperature compensation, so that the matrix M may include temperature compensation parameters as elements, which may be retrieved, for example, from a look-up table.

[0158] The evaluation device 140 is configured to generate at least one item of measurement information by applying the classification model and the transfer function to an item of spectral data. An item of measurement information may be knowledge or evidence providing a qualitative and / or quantitative description of at least one measurement, in particular at least one measurement object. An item of measurement information may include at least one of the physical properties of the measurement object and the chemical properties of the measurement object 118, in particular the chemical composition of the measurement object 118. The physical properties may in particular include optical properties such as at least one absorptivity of the measurement object 118 and / or at least one emissivity of the measurement object 118. The chemical composition may in particular refer to qualitative and / or quantitative information about at least one material that the measurement object 118 comprises. The evaluation device 140 may be configured to perform a classification of the measurement object 118 based on the item of measurement information. The item of measurement information, and thus the measurement object 118, may be assigned to one or more classes, for example classes predetermined by a user. Additionally or alternatively, the items of measurement information may be used to create and / or define at least one class, for example a new material class.

[0159] The transfer function may be updated, in particular recalibrated, by a user of the spectral measurement device 110 or automatically by the spectral measurement device 110. For example, the evaluation device may be further configured to adapt the transfer function, such as when a calibration measurement is performed. In particular, the evaluation device 140 may be configured to update the transfer function accordingly if a change in hardware and / or operating conditions is detected in the calibration measurement. As an example, the calibration measurement may reveal a change in the responsivity of the at least one photodetector 120, for example due to degradation of the at least one pixel 121. The weighting of the pixel 121 may then be adapted in the transfer function, in particular by adapting a corresponding parameter of the transfer function. The spectral measurement device 110 may be equipped with means for automatic self-calibration, for example as described in EP application no. 20211174.6, filed on 2 December 2020, and as described with respect to the optical window 126 of the embodiment of FIG. 1. Additionally or alternatively, instead of using Fresnel reflection of the window material of the optical window 126, the spectral measurement device 110 can be equipped with an integrated diffuse reflection target, which can be used as a calibration target when no sample is present. The spectral measurement device 110 can be configured to perform at least one calibration measurement. In particular, a calibration measurement can be performed when changes in the operating and ambient conditions of the classification-based spectral measurement exceed predefined limits.

[0160] For example, the evaluation device 140 may be configured to perform a calibration of the spectral measurement device 110 by using a first detector signal as generated by one of the photodetectors 120 upon illumination with a first portion 128 of the optical radiation 112 that is guided directly to one of the photodetectors 120 via the optical window 126. The evaluation device 140 is further configured to determine spectral information related to the measurement object 118 by using a second detector signal as further generated by one of the photodetectors 120 upon illumination with a second portion 130 of the optical radiation 112 that is reflected by the measurement object 118 and / or transmitted through the measurement object 118 towards the photodetector 120.

[0161] The evaluation device 140 may be configured to receive the first detector signal and preferably the second detector signal via the interface 146, in particular in a wired or wireless manner. In general, the evaluation device 140 may be part of a data processing device and / or may comprise one or more data processing devices. The evaluation device 140 may comprise one or more additional components, in particular one or more electronic hardware components and / or one or more software components and / or one or more control units. As illustrated here, the evaluation device 140 may further be designed to fully or partially control or drive the spectrum measurement device 110, or components thereof, in particular the photodetector 120, in particular via the interface 146, and / or the radiation source 116, in particular via the further interface 146.

[0162] In the exemplary embodiment shown in FIG. 1, the evaluation device 140 is fully integrated in the housing 114, but an external evaluation unit (not shown here) may also be possible, which may be provided as a separate entity located outside the housing 114, such as part of an electronic communication unit, in particular a smartphone or a tablet. The information determined by the evaluation device 140 may be provided to one or more further devices or to the user in an electronic, visual, acoustic and / or tactile manner. As an example, the information may be displayed using a monitor of the smartphone. Furthermore, the information may be stored in a data storage unit (not shown here), which may be included by the evaluation device 140 or by another storage device, such as a smartphone.

[0163] The evaluation device 140 may be configured to automatically recalibrate the spectral measurement device 110 by using at least one photodetector signal generated by using the dark pixels 124, in particular for continuously or repeatedly recalibrating the spectral measurement device, for example in parallel with performing the classification-based spectroscopic measurement. The photodetector signal generated by using the dark pixels 124 may be independent of the optical radiation reaching the spectral measurement device. In other words, the photodetector signal generated by using the dark pixels 124 may be constant even if the optical radiation reaching the spectral measurement device 110 changes. The photodetector signal generated by using the dark pixels 124 may be suitable for correcting drifts of the photodetector 120, especially drifts over time. The photodetector signal generated by using the dark pixels may be temperature dependent. As an example, the dark pixels 124 may include at least a semiconductor material in which spontaneous formation of free charge carriers can be thermally induced. Thus, the photodetector signal generated by using the dark pixels may be even more suitable for correcting temperature effects on the photodetector 120. The calibration may include correcting drift effects that may occur in the actual measurement due to changes primarily related to the hardware components and / or due to changes affecting the hardware components, specifically the spectral measurement device or parts thereof. The changes may include, specifically, at least one of the following: degradation of at least one of the radiation source 116 or the photodetector 120, specifically the pixels 121; temperature drift of at least one of the radiation source 116 or the photodetector 120; variations in the ambient temperature affecting the spectral measurement device 110; variations in the temperature related to the spectral measurement device 110, i.e., variations in the temperature at which the photodetector 120 and / or the corresponding electronics may operate; mechanical expansion and contraction of at least one component included by the spectral measurement device 110, specifically the mechanical housing, holder and / or optical elements. Further changes may also be possible. Physical processes such as electrochemical processes or relaxation of long-lived traps may induce drift effects.Correcting drift effects can facilitate maintaining the reliability of an item of measurement information, in particular by avoiding drift effects distorting the item of measurement information to such an extent that the results determined by the spectrum measurement device 110 become inconclusive. Periodic calibration of the hardware used may therefore typically be necessary to maintain the reliability of the spectrum measurement device. Automatically calibrating the spectrum measurement device by using at least one photodetector signal generated by using a dark pixel may be particularly consumer-friendly. The photodetector signal generated by using a dark pixel may be suitable for correcting photodetector drift, particularly long-term drift, and also for correcting temperature effects on the photodetector 120.

[0164] 2A and 2B show an arrangement of a plurality of single pixels 121 with wavelength-dependent optical filters 138 with a single radiation source 116. In the perspective view of FIG. 2A, eight photodetectors 120, each with a single pixel 121, are shown in a circular arrangement around one single radiation source 116. Exemplarily, three optical filters 138 configured to transmit specific wavelengths to the photodetectors 120 are shown in FIG. 2A. The photodetectors 120 may be separated from each other so that crosstalk between the individual detectors 120 and stray light is suppressed. In FIG. 2A and 2B, blocking walls 148 made of IR absorbing material are arranged between the photodetectors 120. This is believed to be possible because the photodetectors 120 are physically separated on individual substrates. Thus, radiation at a specified wavelength can be detected more accurately without crosstalk. In the top view of FIG. 2B, the circular array of photodetectors 120 is shown without optical filters 138.

[0165] Figures 3A and 3B show an embodiment where all pixels are twins configured for a particular wavelength and the twin is darkened: Figure 3A shows one active twin pixel 121, and Figure 3B shows a twin with a dark pixel 124.

[0166] Figures 5A and 5B show a further assembly of a photodetector 120 with a radiation source 116. In Figure 5A, the radiation source 116 is centrally located. In Figure 5B, multiple radiation sources 116 can be used arranged around the periphery. In both figures, a shielding wall 148 is shown. Additionally, an optical window 126 is shown.

[0167] As outlined above, the spectral measurement device 110 may be designed specifically for each use case. For example, a textile use case may call for differentiation of different fibers into classes such as cotton, wool, silk, synthetics, etc. Every fiber has a specific characteristic infrared absorption spectrum. It has already been shown in the literature that certain materials have specific absorption bands. For example, cotton absorbs around 2100 nm, wool absorbs around 1724 nm, synthetics around 1662 nm, etc. Figure 4 shows the material specific absorption bands for different fiber classes. Therefore, instead of measuring the entire spectrum with an expensive spectrometer, it may be sufficient to measure these four material specific absorption bands by setting the detector pixels to these sample specific wavelengths, since in this textile use case only certain wavelength components carry information about the fiber class. Four photodetectors 120 may be used, for example photodetectors 1 to 4. The photodetectors 120 may be wavelength tuned using narrow band pass (NBP) optical filters 138 with center wavelengths and full width at half maximum (FWHM) values ​​as shown in the table below: [Table 2]

[0168] For example, a further use case may require classification of plastics, e.g., PET is classified from other plastic types such as PP, PE, etc. In this embodiment, two or three pixels specific to these plastic classes can be used.

[0169] 6 shows a schematic diagram of an exemplary embodiment of a method for measuring optical radiation 112 according to the present invention. The method comprises the following steps: (Reference numeral 150) a step of emitting optical radiation 112 at least partially towards a measurement object 118 by using at least one radiation source 116, said optical radiation 112 being at least partially in a spectral range that is subject to a classification-based spectral measurement; (Reference numeral 152) detecting optical radiation 112 provided by the measurement object 118 and generating at least one optical detector signal by using at least two optical detectors 120; (reference numeral 154) measuring said photodetector signal and generating at least one item of spectral data by using at least one readout device 139; (reference numeral 156) a step of transferring an item of the spectral data to at least one item of input data applicable to a classification model by using a transfer function; (Reference numeral 158) generating at least one item of measurement information by using said classification model; Includes.

[0170] The method comprises the following steps (not shown here): performing at least one calibration measurement when changes in operating conditions and / or ambient conditions of the classification-based spectral measurement exceed predefined limits; fitting the transfer function corresponding to the calibration measurements; Further includes: [Explanation of symbols]

[0171] 110 Spectroscopic Instrument 112 Light Radiation 114 Housing 116 Radiation Source 118 Measuring object 120 Photodetector 121 pixels 123 active pixels 124 dark pixels 126 Optical window 128 Part 1 130 Part 2 134 Optically transparent media 136 Spectral Transfer Elements 138 Optical Filter 139 Readout device 140 Evaluation device 142 processors 144 Memory Storage 146 Interface 148 Blocking Wall 150 Method step a) 152 Method step b) 154 Method step c) 156 Method step d) 158 Method step e)

Claims

1. A spectral measurement device (110) for measuring optical radiation (112) provided by at least one measurement object (118) with at least one classification-based spectral measurement, said spectral measurement device (110) comprising: at least one radiation source (116) configured to emit optical radiation (112) at least partially towards said measurement object (118), said optical radiation (112) being at least partially within a spectral range that is subject to a classification-based spectral measurement; at least two photodetectors (120), each photodetector (120) comprising at least one pixel (121), each pixel (121) being an active pixel (123) or a dark pixel (124), said spectral measurement device (110) comprising at least two active pixels (123), said spectral measurement device (110) comprising at least one dark pixel (124), each active pixel (123) configured to generate at least one photodetector signal dependent on illumination of said active pixel (123), at least two of said active pixels (123) configured to detect optical radiation in at least partially different spectral ranges, each dark pixel (124) configured to generate at least one photodetector signal independent of illumination of said dark pixel (124); at least one readout device (139) configured to measure said photodetector signal and to generate at least one item of spectral data; at least one evaluation device (140) comprising at least one processor (142) and at least one memory storage (144), said memory storage (144) comprising: at least one classification model including a set of distinct classes, each class referring to at least one spectral characteristic, said classification model configured to classify at least one item of input data into said class; at least one transfer function configured to transfer the items of spectral data to items of input data applicable to the classification model; At least one evaluation device (140) configured to store Equipped with The evaluation device (140) is configured to generate at least one item of measurement information by applying the classification model and the transfer function to an item of the spectral data.

2. 2. The spectral measurement device (110) of claim 1, wherein each photodetector (120) includes at most one active pixel (123), the spectral measurement device (110) includes at least one dark pixel (124), and the evaluation device (140) is configured to automatically recalibrate the spectral measurement device (110) by using at least one photodetector signal generated by using the dark pixel (124).

3. The spectral measurement device (110) of claim 1, wherein each active pixel (123) is configured to detect optical radiation (112) having at least one spectral characteristic that references at least one class of the classification model.

4. The spectral measurement device (110) of claim 1, wherein each class refers to at least one material having at least one distinct spectral characteristic or at least one group of materials having similar spectral characteristics.

5. The spectral measurement device (110) of claim 1 , wherein each spectral characteristic refers to at least one predetermined and / or predefined spectral range.

6. The spectral measurement device (110) of claim 1, wherein the classification model is based on at least one algorithm trained on past spectral data acquired by at least one external device, in particular at least one further spectral measurement device (110).

7. 2. The spectral measurement device (110) of claim 1, wherein the transfer function comprises at least one parameterized mathematical function including at least one parameter, and the transfer function takes into account at least one of the hardware of the spectral measurement device (110) and operating conditions of the spectral measurement device (110).

8. The spectral measurement device (110) of claim 1, wherein the items of measurement information include at least one of a physical property of the measurement object (118) and a chemical property of the measurement object (118), in particular a chemical composition of the measurement object (118).

9. The spectral measurement device (110) of claim 1, wherein the evaluation device (140) is configured to perform a classification of the measurement object (118) based on the items of measurement information.

10. 2. The spectral measurement device of claim 1, wherein the spectral measurement device is configured to perform at least one calibration measurement, and the evaluation device is further configured to adapt the transfer function when a calibration measurement is performed.

11. 2. The spectral measurement device of claim 1, wherein at least one of the photodetectors includes at least one photoconductive material selected from at least one of PbS, PbSe, Ge, InGaAs, InSb, or HgCdTe.

12. 2. The spectral measurement device (110) of claim 1, wherein the spectral measurement device (110) comprises at least one communication interface configured to transmit data from and / or to the evaluation device (140), in particular for adapting the classification model and / or the transfer function.

13. 2. The spectral measurement device (110) of claim 1, comprising at least one sample interface configured to at least partially direct optical radiation (112) emitted by the radiation source (116) toward the optical detector (120) in a predefined spectrum.

14. A method for measuring optical radiation (112) provided by at least one measurement object (118) by performing a classification-based spectral measurement by using a spectral measurement device (110) according to any one of claims 1 to 13, said method comprising: emitting optical radiation (112) at least partially towards the measurement object by using at least one radiation source (116), the optical radiation (112) being at least partially in a spectral range that is subject to a spectral measurement based on the classification; detecting the optical radiation (112) provided by the measurement object (118) and generating at least one optical detector signal by using at least two optical detectors (120); measuring said photodetector signal and generating at least one item of spectral data by using at least one readout device (139); transferring said item of spectral data to at least one item of input data applicable to a classification model by using a transfer function; generating at least one item of measurement information by using said classification model; A method comprising:

15. A computer program comprising instructions, when the program is executed by a spectrum measuring device (110) according to any one of claims 1 to 13 referring to the spectrum measuring device (110), causing the spectrum measuring device (110) to perform the method according to claim 14 referring to the method.

16. 14. Use of the spectral measurement device (110) according to any one of claims 1 to 13 with reference to the spectral measurement device (110), for a purpose selected from the group consisting of: infrared detection applications; spectroscopy applications; exhaust gas monitoring applications; combustion process monitoring applications; pollution monitoring applications; industrial process monitoring applications; mixing or blending process monitoring applications; chemical process monitoring applications; food treatment process monitoring applications; food preparation process monitoring applications; water quality monitoring applications; air quality monitoring applications; quality control applications; temperature control applications; motion control applications; exhaust control applications; gas detection applications; gas analysis applications; motion sensing applications; chemical detection applications; mobile applications; medical applications; mobile spectroscopy applications; food analysis applications; agricultural applications, in particular soil, silage, feed, crop or agricultural product characterization, plant health monitoring; plastic identification and / or recycling applications; textile identification and / or recycling applications.

Images