METHOD AND DEVICE FOR DETERMINING WHETHER AN OIL FRUIT, A NUT, IN PARTICULAR A HAZELNUT OR A SEED IS ROTTEN

Description

[0001] The invention relates to a method for determining whether an oilseed, a nut, in particular a hazelnut or a seed is rotten, according to claim 1.

[0002] The invention further relates to a device for determining whether an oilseed, a nut, in particular a hazelnut or a seed is rotten, according to claim 12. Technological background

[0003] The detection and subsequent sorting of bulk materials using photosensors is a common method applied on an industrial scale. In such methods, known in the prior art, seeds are individually examined spectroscopically by irradiating them with a light source. An absorption or reflection spectrum is then recorded by a photosensor. A computer unit subsequently analyzes the absorption or reflection spectrum of each seed within a region of interest and calculates the concentration of a specific component of the seed using a calibration curve.

[0004] The detection of various constituents within individual elements of a bulk material is of interest, for example, in the food industry, to differentiate between spoiled and unspoiled elements. According to current technology, such methods typically operate in the near-infrared range. To enable the use of these methods in production facilities, the photosensors employed must have high frame rates, usually 500 Hz or more. This ensures high throughput while simultaneously guaranteeing reliable analysis of the constituents of each individual element examined. Traditionally, the data acquired by the photosensors are analyzed using common statistical classification methods such as partial least squares, principal component regression, or similar techniques.This qualitative analysis yields excellent results when there are clear differences between spoiled and unspoiled elements in the absorption or reflection spectrum. Hazelnuts are processed whole in many products. Hazelnuts that appear perfectly fine on the outside have often already begun to rot inside. Therefore, they cannot be distinguished using conventional optical methods in the visible spectrum. The causes are varied and include improper storage, insect infestation, or mold growth. Often, the rotting process is also associated with the formation of dangerous molds (toxins, e.g., aflatoxin). Mold develops due to a mold fungus: In microbiology, mold fungi are a systematically heterogeneous group of filamentous fungi, the majority of which belong to the taxonomic groups of Ascomycetes and Zygomycetes. Mold fungi such as... Aspergillus flavusThey occur in soil, decaying vegetation, and hay and grain exposed to microbial spoilage. A mold fungus attacks the fruit from the outside and settles on the surface.

[0005] In ecology and thanatology, putrefaction is defined as the decomposition of biotic substances by microorganisms under oxygen-deficient conditions. Often, the term "putrefaction" is specifically used when decomposition is accompanied by an unpleasant odor. Putrefaction is a natural form of fermentation and is referred to here as rot. The development of putrefaction differs significantly from mold. In addition to altered taste, it can also lead to health problems. In any case, rotten hazelnuts, whether raw or roasted, must be removed from the product. However, since this is not a simple yes / no decision, but rather a gradual quality criterion, a precise assessment and evaluation of the degree of putrefaction is necessary.

[0006] Prior art patent CN 113420614 A discloses a method for identifying moldy peanuts, in particular a method for identifying moldy peanuts based on a near-infrared hyperspectral image using a deep learning algorithm with a neural network. The neural network described therein comprises 265 neurons at the input and two neurons at the output. Furthermore, a characteristic wavelength of 1343 nm is mentioned, which is used to perform pre-segmentation. This allows for the detection of the mold product aflatoxin.A distribution map of peanut mold information is generated, whereby, by the number of moldy pixels at the detector and a threshold (β), each peanut particle in the distribution map of mold information is recorded in order to identify a moldy peanut and to generate an image of the identification result of moldy peanuts.

[0007] A disadvantage of this known solution is that mold growth occurs on the peanut surface, meaning only surface information of the peanut is assessed. While the apparently necessary, detailed deep learning algorithm, due to its parameterization, may achieve the desired success in detecting mold on peanuts, it is unsuitable for detecting rot or decay at an early stage.

[0008] Prior art includes AT 519918 A1 and WO 2018 / 191 768 A1. These disclose a method for detecting rancidity in oilseeds, seeds, and nuts. Rancidity is the state into which fats and other lipids degrade through oxidation or by fat-splitting enzymes (lipases). The spoilage of vegetable and animal fats, which can be detected in its early stages by changes in odor and taste (rancidity), is due, on the one hand, to hydrolysis and the associated breakdown of longer-chain fats in the case of water-containing fats, and, on the other hand, to the action of atmospheric oxygen (oxidation). The patent application comprises irradiating a sample of an oilseed, nut, or seed with a light source and recording the absorption or reflection spectrum of the reflected or transmitted light.The absorption or reflection spectrum is then compared with an external chemical analysis of the sample to establish a correlation between the absorption or reflection spectrum and rancidity of the sample.

[0009] A disadvantage of known prior art methods is that only the rancidity of the oilseed or nut can be determined, and that as soon as a marker is detected in the absorption or reflection spectrum that classifies a product as defective, it is immediately rejected. However, if the differences between the spectrum of a good product and a defective product are very small, this small difference is amplified so much in the external process that the signal-to-noise ratio increases significantly, and the uncertainty in distinguishing between good and defective products is greatly amplified. Furthermore, the differences in the spectrum are determined using a reference sample of good and defective specimens. However, since these are natural products with spectral variation, this comparison is subject to a high degree of uncertainty and can fluctuate in each batch. Contamination or...Product defects are also often unevenly distributed throughout the product. The result of misclassification is an inadequate separation of good and spoiled food.

[0010] Prior art document CN104544334 A discloses the performance of color sorting by a color sorter to sort out mealy and rotten nuts.

[0011] Conventional methods for determining the quality of oilseeds, nuts (especially hazelnuts), and seeds in the food industry result in a large amount of waste, even though at least some of these rejected products could still be used or processed further. At the same time, excessive sorting also leads to economic disadvantages. Description of the invention

[0012] One object of the invention is to avoid at least one of the disadvantages of the prior art. In particular, this object is achieved by providing a method and a device for determining whether an oilseed, a nut, especially a hazelnut, or a seed is rotten, in order to reduce the reject rate in the determination process compared to known methods. This is particularly critical when products contain substances that cause health problems for consumers. In any case, if the quality of the product is impaired, an impairment of taste is to be expected. This is particularly the case in hazelnut processing.

[0013] This problem is solved by the features of the independent patent claims. Advantageous developments are set out in the figures and in the dependent patent claims.

[0014] The inventive method for determining whether an oilseed, a nut, in particular a hazelnut or a seed is rotten, comprises at least the following calibration steps: a) Irradiating a sample of an oilseed, a nut, in particular a hazelnut, or a seed with a calibration light source, b) Acquiring an absorption or reflection spectrum in a wavelength range of 300–2500 nm using a calibration photosensor, c) Assigning at least one characteristic wavelength or at least one characteristic wavelength range of the acquired absorption or reflection spectrum of the sample to a degree of putrefaction, d) Repeating the preceding steps for a representative number of samples and establishing a correlation between the degree of putrefaction and the assigned wavelengths or wavelength ranges of the acquired absorption or reflection spectra.

[0015] The calibration steps establish a correlation between a value of the absorption or reflection spectrum at a characteristic wavelength, or at least within a characteristic wavelength range, and the degree of putrefaction of the sample. A correlation describes a relationship between two or more characteristics, states, or functions, with a causal correlation being the preferred approach here.

[0016] This makes it possible to determine the degree of rot of an oilseed, a nut, especially a hazelnut, or a seed purely by evaluating the absorption or reflection spectrum, without the need for a separate additional analysis, such as a laboratory analysis.

[0017] Preferably, the calibration steps are carried out in the aforementioned order to enable reproducible calibration, especially for hazelnuts.

[0018] Furthermore, the method according to the invention comprises the following detection steps: e) Irradiating an oilseed, a nut, in particular a hazelnut, or a seed with a detection light source; f) Acquiring absorption or reflection spectra in a wavelength range of 300–2500 nm using a detection photosensor, wherein the detection photosensor has a plurality of pixels and each pixel acquires an absorption or reflection spectrum; g) Assigning a degree of rot to each pixel of the detection photosensor by applying the correlation according to step d) to the absorption or reflection spectra acquired by the pixels of the detection photosensor; h) Classifying the oilseed, the nut, in particular a hazelnut, or the seed as rotten if at least a certain number of pixels exhibit a degree of rot exceeding a first threshold, and / or if a degree of rot assigned to at least one pixel exceeds a second threshold.

[0019] The detection steps enable a reproducible evaluation of recorded absorption or reflection spectra based on the previously performed calibration steps.

[0020] The detection steps are preferably carried out in the aforementioned order to enable reproducible detection of rot, especially for hazelnuts.

[0021] According to the invention, the absorption or reflection spectra for each individual pixel of the detection photosensor are recorded, and a degree of rot is assigned to each of these spectra by means of the correlation established in the calibration steps. This achieves the advantage that a single absorption or reflection spectrum for each irradiated oilseed, nut, especially hazelnut, or seed is not used for evaluation to determine whether the oilseed, nut, especially hazelnut, or seed in question is rotten. An additional criterion can thus be introduced, which consists of a certain number of pixels exhibiting a degree of rot that exceeds a first threshold value. The detection photosensor has a plurality of pixels and records an absorption or reflection spectrum for each pixel.The first threshold value is therefore checked for each pixel on the detection photosensor.

[0022] The measured values ​​or spectra per pixel at the detection photosensor are defined as the degree of inefficiency, for example, in a range of 0-100%. Subsequently, sorting can be carried out based on the first or second threshold value, for example, in a sorting system.

[0023] Alternatively or additionally, the oilseed, nut, or seed in question can also be rejected if the degree of putrefaction assigned to at least one pixel exceeds a second threshold. By defining the first threshold and the number of pixels whose assigned degree of putrefaction must exceed this first threshold, as well as the second threshold, the inventive method makes it possible to take into account inhomogeneities in the putrefaction process in the agricultural natural products examined using the inventive method. This allows for a significantly more targeted sorting into good and bad products than was previously possible with methods according to the prior art.

[0024] The method according to the invention makes it possible to determine the putrefaction of oilseeds, nuts, in particular hazelnuts, or seeds by means of a spatially resolved determination of the degree of putrefaction. Furthermore, the first and second threshold values ​​allow for the inclusion of an inhomogeneous distribution of fatty acid degradation products in the oilseed, nut, in particular hazelnut, or seed in the assessment of putrefaction. This enables an improved differentiation between, for example, a non-putrefied hazelnut and a putrefied hazelnut.

[0025] Preferably, the sample of oilseed, nut (especially hazelnut), or seed moves past the respective light source in step a) and / or step e). This makes the aforementioned method applicable to moving oilseeds, thus simplifying the sorting of spoiled and good oilseeds during the process. Large mass flows of oilseeds, such as 6 t / h, can be sorted in a very short time using the aforementioned method.

[0026] Preferably, prior to step b), the light reflected and / or transmitted by the sample is projected onto a calibration photosensor. This allows for improved recording of the spectra.

[0027] Preferably, a wavelength range of 900-1700 nm is used in step b). This improves the measurability of the infrared range of the light and the associated information for calibrating the degree of laziness.

[0028] Preferably, after step b), a degree of putrefaction is determined based on the content of at least one fatty acid degradation product in the sample. Thus, fatty acid degradation products are searched for within the spectral range used, in order to provide reproducible detection of putrefied oilseeds.

[0029] In particular, after step b), a degree of putrefaction is determined based on the content of at least one fatty acid degradation product in the sample and additionally based on the content of acetic acid in the sample. The content of the at least one fatty acid degradation product and the content of acetic acid are mathematically linked in the evaluation, for example by averaging, calculating the median, or adding them together. Alternatively or additionally, the contents can be weighted. This increases the accuracy of the classification in the method according to the invention.

[0030] Preferably, the at least one fatty acid degradation product comprises at least one component from the group consisting of butyrolactone, diacetyl, 2-methylbutanal, 3-methylbutanal, acetylacetone, filbertone, or 2,3-butanediol. A model can be provided that is not limited to a single chemical substance but rather to those components that exhibit a significant concentration difference between good and spoiled oilseeds, nuts (especially hazelnuts), or seeds. This means that, within the relevant spectral range, the quantitative degree of flavor-altering substances can be automatically inferred from the relative or absolute amplitudes in the spectrum of several spoiled components or substances measured. For example, the values ​​can be normalized, and then a derivation can be applied to the spectra for comparison.

[0031] Preferably, the degree of putrefaction is determined based on the levels of at least two fatty acid degradation products in the sample. For example, the two fatty acid degradation products that exhibit the greatest concentration difference between putrefied and good oilseed are used. The levels are mathematically correlated in the evaluation, for example, by averaging, calculating the median, or summing. Alternatively or additionally, the levels can be weighted. By using at least two constituents characteristic of the putrefaction process to determine the degree of putrefaction, the accuracy of the method according to the invention is further increased.

[0032] Preferably, several components of fatty acid degradation products are sought that exhibit the greatest concentration difference between good and spoiled oilseeds, nuts (especially hazelnuts), or seeds. The model is then broken down, for example, to five components that show the greatest concentration difference between spectra of good and spoiled oilseeds, nuts (especially hazelnuts), or seeds. This means that, within the relevant spectral range, the quantitative degree of flavor-altering substances can be automatically inferred from the measured values ​​of several substances and their absolute amplitudes in the spectrum.

[0033] Preferably, the correlation is at least a correlation function, or at least an index table, or a table of values, or at least comprises a comparison of spectral information that includes at least the degree of putrefaction and the corresponding wavelengths or wavelength ranges of the recorded absorption or reflection spectra. A correlation can comprise a correlation function, an index table, a table of values, or the like, such that a causal relationship between the degree of putrefaction and the corresponding wavelengths or wavelength ranges of the recorded absorption or reflection spectra can be established. A correlation function enables a continuous relationship between the degree of putrefaction and the corresponding wavelengths or wavelength ranges of the recorded absorption or reflection spectra, and thus a complete and accurate causal relationship.Although index tables or value tables do not include a continuous causal relationship between the degree of rot and the assigned wavelengths or wavelength ranges of the recorded absorption or reflection spectra, this is sufficient for some oilseeds to enable a reproducible sorting of rotten and good oilseeds.

[0034] Preferably, prior to step e), the light reflected and / or transmitted by the oilseed, nut (especially hazelnut), or seed is projected onto a detection photosensor. This allows for an improved correlation of the measured values ​​with the degree of rot.

[0035] Preferably, a wavelength range of 900-1700 nm is used in step e). This improves the measurability of the infrared range of the light and the associated information for detecting the degree of decay.

[0036] According to a preferred embodiment of the inventive method, the assignment of at least one characteristic wavelength or at least one characteristic wavelength range of the recorded absorption or reflection spectrum of the sample to the degree of putrefaction is carried out by means of at least one mean value or median, bandwidth, or individual frequency bands of the recorded absorption or reflection spectra. This achieves the advantage of establishing a targeted correspondence between the recorded absorption or reflection spectra and the degree of putrefaction.

[0037] Preferably, the correlation in an operating mode characterized by the detection steps is used to assign a corresponding measured value or reference value for the degree of decay to newly recorded spectra.

[0038] Preferably, determining the concentration of at least one fatty acid degradation product in the sample involves performing a measurement in the laboratory, specifically gas chromatography followed by mass spectrometry. Crucially, the laboratory analysis of the chemical substances must clearly distinguish between the desired and undesired products. This distinction is readily apparent on a scatter plot of the chemical substances used over the measured samples, such as hazelnut samples. This separation is a necessary but not sufficient condition for a sound quantitative model. Whether a substance can then be quantitatively determined based on its NIR spectrum depends on the position of additional chemical substances, which may well be capable of masking the absorption band.

[0039] Preferably, each pixel corresponds to a reference value based on calibration, determined by the correlation between spectral data and the measurement laboratory. The resulting correlation, i.e., the spectral information that can be related to the information in the reference dataset, is then used in an operating mode characterized by the detection steps to assign a corresponding measurement value for the degree of decay to newly acquired spectra.

[0040] According to a preferred embodiment of the method according to the invention, a single sensor is used as both the calibration photosensor and the detection photosensor. This offers the advantage that only a single sensor is required for both the calibration and detection steps. Additionally, a single light source can be used as both the calibration and detection light sources. This further reduces the cost of using the method according to the invention.

[0041] Preferably, the absorption or reflection spectra are acquired by the calibration photosensor and / or the detection photosensor using hyperspectral imaging. This allows for the acquisition of a particularly wide wavelength range. Hyperspectral cameras are preferably used for this purpose. In particular, the absorption or reflection spectra are acquired by the calibration photosensor and / or the detection photosensor using a color camera. This allows for the acquisition of an additional range of 380–780 nm. For example, discolored oilseeds can be detected with the color camera and easily sorted out. In addition to rotten oilseeds, discolored oilseeds can also be sorted out, thus improving the sorting quality. Discoloration can indicate further undesirable quality reductions in oilseeds. It is particularly preferred if the measured information from the color camera is evaluated with the AI ​​module described below.This allows the desired recognition rate to be increased to 99%.

[0042] According to the preferred embodiment of the method according to the invention, determining the content of at least one fatty acid degradation product in the sample comprises performing gas chromatography followed by mass spectrometric analysis. This allows for a particularly precise determination of the fatty acid degradation product content.

[0043] The device according to the invention for determining whether an oilseed, a nut, in particular a hazelnut, or a seed is rotten comprises a calibration light source, a detection light source, a calibration photosensor and a detection photosensor and is configured to carry out at least one method disclosed herein according to the invention.

[0044] Preferably, the device comprises a sorting unit, wherein the sorting unit is configured to separate an oilseed, a nut, in particular a hazelnut, or a seed from a product stream if the oilseed, the nut, in particular the hazelnut, or the seed is classified as rotten. This achieves the advantage that the process according to the invention can be implemented on a large industrial scale.

[0045] A preferred embodiment of the aforementioned method for determining whether an oilseed, a nut, in particular a hazelnut or a seed is rotten, comprises the following calibration steps: Irradiating a sample of an oilseed, a nut, in particular a hazelnut, or a seed with a calibration light source, projecting the light reflected and / or transmitted by the sample onto a calibration photosensor (4), acquiring an absorption or reflection spectrum in a wavelength range of 300-2500 nm, preferably 900-1700 nm, by the calibration photosensor, determining the content of at least one fatty acid degradation product in the sample, determining a degree of putrefaction based on the content of the at least one fatty acid degradation product in the sample, and assigning at least one characteristic wavelength or at least one characteristic wavelength range of the acquired absorption or reflection spectrum of the sample to the degree of putrefaction.Repeating the preceding steps for a representative number of samples and forming a correlation function between the degree of rot and the associated wavelengths or wavelength ranges of the detected absorption or reflection spectra, characterized in that the method comprises the detection steps of irradiating an oilseed, a nut, in particular a hazelnut, or a seed with a detection light source, projecting the light reflected and / or transmitted by the oilseed, the nut, in particular the hazelnut, or the seed onto a detection photosensor, and detecting absorption or reflection spectra in a wavelength range of 300-2500 nm, preferably 900-1700 nm, by the detection photosensor, wherein the detection photosensor has a plurality of pixels and detects an absorption or reflection spectrum for each pixel.Assigning a degree of rot to each pixel of the detection photosensor by applying the correlation function to the absorption or reflection spectra acquired by the pixels of the detection photosensor, classifying the oilseed, nut, especially hazelnut, or seed as rotten if at least a certain number of pixels exhibit a degree of rot exceeding a first threshold, and / or if a degree of rot assigned to at least one pixel exceeds a second threshold.

[0046] A single sensor can be used for both calibration and detection purposes.

[0047] Furthermore, a light source can be used as a calibration light source and as a detection light source.

[0048] In particular, the degree of putrefaction is determined based on the levels of at least two fatty acid degradation products in the sample.

[0049] In particular, the assignment of at least one characteristic wavelength or at least one characteristic wavelength range of the recorded absorption or reflection spectrum of the sample to the degree of putrefaction is carried out by means of at least one of a mean value, a bandwidth or individual frequency bands of the recorded absorption or reflection spectra.

[0050] Preferably, the absorption or reflection spectra are acquired by the calibration photosensor and / or the detection photosensor using hyperspectral acquisition.

[0051] In particular, determining the content of at least one fatty acid degradation product in the sample includes performing gas chromatography and subsequent mass spectrometric analysis.

[0052] A preferred embodiment of the device for determining whether an oilseed, a nut, in particular a hazelnut, or a seed is rotten comprises a calibration light source, a detection light source, a calibration photosensor and a detection photosensor, and is configured to carry out one of the aforementioned methods.

[0053] Preferably the device comprises a sorting unit, wherein the sorting unit is configured to separate an oilseed, a nut, in particular a hazelnut, or a seed from a product stream if the oilseed, the nut, in particular the hazelnut, or the seed is classified as rotten.

[0054] Preferably, the device includes a control unit. The control unit can control at least one component from the group consisting of a calibration photosensor, a detection photosensor, a calibration light source, and a detection light source, so that at least one previously described procedure can be carried out automatically and reproducibly. In particular, the control unit causes an output device to output at least one measured value. For example, a display is provided as the output device, on which at least indicative information about the putrefaction of the oilseed is displayed. For example, a distribution function or statistical data for the putrefied oilseed present in the product stream or mass stream is displayed. This allows for improved process monitoring.

[0055] In particular, the control unit controls the sorting unit. The control unit provides control data that regulates the sorting unit. For example, the sorting unit can include a pneumatic deflection unit connected to the detection photosensor. Using the control data, the pneumatic deflection unit separates the oilseeds, nuts (especially hazelnuts), or seeds classified as rotten from a substantially continuous product stream. The force, and especially the direction, of the compressed air in the pneumatic deflection unit can be controlled.

[0056] Preferably, the sorting unit operates with at least two adjustable thresholds, wherein a first adjustable threshold includes a quantity of substance for a pixel and the second adjustable threshold includes an inhomogeneous distribution of the substance, by taking into account the number of pixels with the attribute: putrid for sorting purposes.

[0057] Preferably, the device includes a processing unit that evaluates spectral data to classify the oilseeds. The spectral data can be measurement data and can be used to generate control data for the control unit, which is then used, in particular, to control the sorting unit.

[0058] In particular, the computing unit is designed to calculate at least one of the two threshold values ​​based on the measurement data or spectral data and to generate control data for the control device from this.

[0059] Preferably, an AI (artificial intelligence) module is present, which is connected to the processing unit. The solution described above requires specialized knowledge of the absorption bands of oilseeds and fatty acid degradation products in the infrared range in order to select the correct spectral range and thus achieve optimal calibration for distinguishing between good and spoiled oilseeds. With the help of deep learning algorithms and neural networks, this specialized chemical-physical knowledge can be dispensed with. Historical measurement data or spectra of predominantly spoiled and predominantly good samples can be provided to the neural network as training data. This neural network automatically finds the optimal range of interest in the measurement data or spectra. This further improves the calibration using the acquired spectral data, especially the hyperspectral data."Poor" training data is automatically weighted less than more meaningful data. Poor training data is data where the good and bad ranges within an oilseed are less clearly distinguishable, making it impossible to create a meaningful training model. This method can further reduce the differentiation, and thus the so-called erroneous parameter defined as rejects, by a factor of 5 to 10. Instead of approximately 5% incorrect rejects in the sorting unit, this can be reduced to less than 1%. With this trained model, it is then possible to predict whether the rejects will fall within the desired range. The AI ​​module can therefore provide essential parameters for assessing rotten oilseeds.

[0060] The AI ​​used here requires no pre-segmentation of the spectra, nor are specific wavelengths passed to the neural network as preferred wavelengths. This allows the AI ​​to learn without any prior information. This has the advantage that the method can respond more flexibly to different putrefaction products or fatty acid degradation products. The trained AI independently selects the best value range for optimal selection results.

[0061] In general, the AI ​​module serves to build a statistical model based on training data, which is then tested using test data before finally being applied to the data in the ongoing product stream. Among the available algorithms are supervised machine learning algorithms, where a training dataset is used to train a model, which is then applied to further evaluation data to calculate a classification. One approach to training such models is deep learning (artificial neural networks), where multiple layers of artificial neurons link the input variables (feature vector) with the output variable (classification, regression, etc.).In addition to numerous other machine learning methods, Random Forest algorithms (randomized decision trees) or Support Vector Machines (estimation using support vectors in the vector space of feature vectors) can also be used, especially to limit the computational effort.

[0062] A computer program product according to the invention comprises program instructions configured to execute at least one of the aforementioned methods. The computer program product may include instruction data, computed data, and parameters, as previously described.

[0063] A computer-readable medium according to the invention comprises at least one computer program product which, when executed by at least one computing unit, causes the latter to carry out at least one of the aforementioned methods. The computer-readable medium may include control data, measurement data, calculated data, measured values, and parameters, as described above.

[0064] The inventive method and device, as well as preferred and alternative embodiments, will be explained in more detail below with reference to the figures.

[0065] Further advantages, features and details of the invention will become apparent from the following description, in which exemplary embodiments of the invention are described with reference to the drawings.

[0066] The list of reference numerals, like the technical content of the patent claims and figures, forms part of the disclosure. The figures are described coherently and comprehensively. Identical reference numerals denote identical components; reference numerals with different indices indicate functionally identical or similar components.

[0067] The invention is explained in more detail with reference to exemplary embodiments in the following figures. The list of reference numerals forms part of the disclosure.

[0068] Positional references, such as "top", "bottom", "right" or "left", refer to the corresponding representations and are not to be understood as restrictive.

[0069] Although the invention is illustrated and described in detail by means of the figures and the accompanying description, this illustration and detailed description are to be understood as illustrative and exemplary and not as limiting the invention. It is understood that those skilled in the art may make modifications and adaptations without departing from the scope of the following claims. In particular, the invention also includes embodiments with any combination of features mentioned or shown above with regard to various aspects and / or embodiments.

[0070] The invention also includes individual features shown in the figures, even if they are shown there in conjunction with other features and / or are not mentioned above. Furthermore, the term "comprises" and derivatives thereof does not exclude other elements or steps. Likewise, the indefinite article "a" or "an" and derivatives thereof does not exclude a plurality. The functions of several features listed in the claims can be fulfilled by a single unit. The terms "essentially," "approximately," "about," and the like, in conjunction with a property or value, also define precisely that property or value. All reference numerals in the claims are not to be understood as limiting the scope of the claims.Terms such as "first" or "second" serve only to distinguish subsequent nouns and do not define any order or evaluation of the subsequent nouns. Character description

[0071] The figures are described in a coherent and comprehensive manner. Identical reference symbols indicate identical components. They show Fig. 1 : a device for determining whether an oilseed, a nut, in particular a hazelnut, or a seed is rotten, Fig. 2 : an alternative embodiment of the device according to the invention, Fig. 3 : another alternative embodiment of the device according to the invention, and Fig. 4 a standardized concentration of fatty acid degradation products in rotten hazelnuts and those fatty acid degradation products with the greatest differences between a rotten hazelnut (Q) and a non-rotten hazelnut (FF). Implementation of the invention

[0072] Figure 1 Figure 1 shows a device 1 according to the invention for determining whether an oilseed, a nut, in particular a hazelnut 2, or a seed is rotten, which is configured to carry out a method according to the invention, in a preferred embodiment. The device 1 according to the invention comprises a calibration light source 3, a detection light source 3', a calibration photosensor 4, and a detection photosensor 4'. The calibration photosensor 4 and the detection photosensor 4' are arranged in the Figure 1 In the illustrated embodiment of the device 1 according to the invention, it is designed as a common photosensor, but it can also be implemented as shown in Figure 3The inventive method for determining whether an oilseed, a nut, in particular a hazelnut 2, or a seed is rotten comprises a series of calibration steps and a series of detection steps. The calibration steps of the inventive method include irradiating a sample of the oilseed, the nut, in particular the hazelnut 2, or the seed with the calibration light source 3. The light reflected and / or transmitted by the sample is subsequently projected onto the calibration photosensor 4, wherein the calibration photosensor 4 detects an absorption or reflection spectrum in a wavelength range of 300–2500 nm, preferably 900–1700 nm.Furthermore, the calibration steps include determining the content of at least one fatty acid degradation product in the sample, establishing a degree of putrefaction based on the content of the at least one fatty acid degradation product in the sample, and assigning at least one characteristic wavelength or at least one characteristic wavelength range of the recorded absorption or reflection spectrum of the sample to the degree of putrefaction. These preceding steps are repeated in the inventive method for a representative number of samples, and a correlation function is established between the degree of putrefaction and the assigned wavelengths or wavelength ranges of the recorded absorption or reflection spectra. The process of putrefaction, for example of a hazelnut, is associated with the formation or degradation of fatty acids.The method according to the invention enables, by means of calibration steps, the correlation of the content of at least one fatty acid degradation product in the sample with a degree of putrefaction, which allows an assessment of the quality of the oilseed, the nut, in particular hazelnut 2, or the seed. This degree of putrefaction is also assigned to a characteristic wavelength or at least a characteristic wavelength range of the recorded absorption or reflection spectrum of the sample, thereby enabling a conclusion to be drawn from the absorption or reflection spectrum to the degree of putrefaction. By repeating the process for a representative number of samples, for example, a correlation function is generated, thus providing a direct inference of the degree of putrefaction via the correlation function by means of an evaluation of the absorption or reflection spectrum.A conventional oilseed, a nut, in particular a hazelnut 2, or a seed can be used as a sample. The device 1 comprises a control unit 7, which controls at least one component from the group consisting of a calibration photosensor, a detection photosensor, a calibration light source, and a detection light source, so that at least one previously described procedure can be carried out automatically and reproducibly. The control unit 7 controls the sorting unit 5. The control unit 7 has control data that controls the sorting unit 5. The device 1 comprises a computing unit 8, which evaluates spectral data to classify the oilseeds. The spectral data can be measurement data and can be used to generate control data for the control unit 7.

[0073] This correlation function can subsequently be applied within the detection steps of the method according to the invention. These detection steps include irradiating an oilseed, a nut, in particular a hazelnut 2, or a seed with the detection light source 3'. The same light source can be used as both the calibration light source 3 and the detection light source 3', which is why in the Figure 1 The device 1 shown according to the invention is provided with only one light source.

[0074] In an alternative or supplementary embodiment, an index table or a table of values ​​can be used, whereby a direct inference to the degree of laziness is provided by means of an evaluation of the absorption or reflection spectrum via the index table or the table of values.

[0075] According to the in Figure 2In the embodiment of the device 1 according to the invention shown, separate light sources can also be used as the calibration light source 3 and as the detection light source 3'. According to the Figure 3 In the illustrated embodiment, separate light sources are used as the calibration light source 3 and as the detection light source 3', and separate photosensors are used as the calibration photosensor 4 and as the detection photosensor 4'. The detection steps further include projecting the light reflected and / or transmitted by the oilseed, the nut, in particular the hazelnut 2, or the seed onto the detection photosensor 4'. As explained with regard to the calibration light source 3 and the detection light source 3', a single sensor can also be used as both the calibration photosensor 4 and the detection photosensor 4'. This reduces the complexity of the device 1 according to the invention. The device 1 according to Figure 2It also includes a computing unit and control device (not shown), as described previously.

[0076] According to the in Figure 3 In the illustrated embodiment of the device 1 according to the invention, different sensors can be used as the calibration photosensor 4 and as the detection photosensor 4'. According to this embodiment, both the detection photosensor 4' and the calibration photosensor 4 can operate reflectively. This means that light reflected from the sample is projected onto the calibration photosensor 4, and the light reflected from the seed 2 is projected onto the detection photosensor 4'. As shown in the Figures 1 and 2However, the calibration photosensor 4 and / or the detection photosensor 4' can also operate transmittively, wherein light transmitted by the sample is projected onto the calibration photosensor 4 and / or the light transmitted by the seed 2 is projected onto the detection photosensor 4'. For the purposes of this invention, an oilseed, a nut, in particular a hazelnut 2, or a seed is understood to mean both a roasted and an unroasted oilseed, nut, in particular a hazelnut, or seed.

[0077] As part of the detection steps, absorption or reflection spectra in a wavelength range of 300-2500 nm, preferably 900-1700 nm, are also acquired by the detection photosensor 4'. The detection photosensor 4' of the device 1 according to the invention for carrying out the method according to the invention has a plurality of pixels and acquires an absorption or reflection spectrum for each pixel. Subsequently, a degree of laziness is assigned to each pixel of the detection photosensor 4' by applying the correlation function to the absorption or reflection spectra acquired by the pixels of the detection photosensor 4'.Furthermore, an oilseed, nut, in particular hazelnut 2 or seed, is classified as rotten if at least a certain number of pixels exhibit a degree of rot exceeding a first threshold, and / or if a degree of rot associated with at least one pixel exceeds a second threshold. The device 1 according to . Figure 3 also includes a computing unit 8 and control unit 7 as well as an AI module 9, wherein the AI ​​module 9 is also included in the devices 1 according to Figure 2 or 3 The AI ​​module may be present. It is connected to or integrated into the processing unit. With the help of the AI ​​module, the differentiation, and thus the number of parameters erroneously defined as rejects, can be reduced by a factor of 5 to 10. Instead of approximately 5% incorrect rejects in sorting unit 5, this can be reduced to less than 1%.

[0078] The method according to the invention makes it possible to determine the putrefaction of the oilseed, the nut, in particular the hazelnut 2, or the seed by means of a spatially resolved determination of the degree of putrefaction. Furthermore, the first and second threshold values ​​allow for the inclusion of an inhomogeneous distribution of fatty acid degradation products in the oilseed, the nut, in particular the hazelnut 2, or the seed in the assessment of putrefaction. This enables an improved differentiation between, for example, a non-putrefied hazelnut 2 and a putrefied hazelnut 2.

[0079] It is known that the process of decay, for example in hazelnuts, is associated with the formation and / or degradation of fatty acids. The standard reference analysis for determining fatty acids or their volatile degradation products is chromatographic determination using gas chromatography (GCC) followed by mass spectrometry. However, this invasive method can only determine either a homogenized composite sample or the content of individual hazelnuts; it cannot selectively identify and remove individual hazelnuts in real time. FTIR analysis should also be mentioned here as a non-invasive standard laboratory method. Based on an interferogram of the sample surface or the homogenized sample, a spectrum is calculated using Fourier transformation. However, this method is also unsuitable for sorting large mass flows, such as 6 t / h, compared to the method according to the invention.During harvesting, but at the latest during the processing, large quantities must now be sorted in such a way as to rule out any health risks and to largely avoid changes in taste. Even before further processing takes place, the method according to the invention enables the automated and high-throughput detection of oilseeds, nuts, especially hazelnuts, or seeds that are altered in taste, toxic, or rotten.

[0080] During the hazelnut harvest, but at the latest during the processing, the large quantities of hazelnuts must be sorted in such a way as to rule out any health risks and to largely prevent changes in taste. Even before the hazelnuts are further processed, for example into chocolate or snacks, any hazelnuts with altered taste, toxicogenic, or rotten flavors should be automatically eliminated at a high throughput rate.

[0081] Since the flavor-altering substances, such as free fatty acids and their degradation products, are spatially inhomogeneously distributed, for example, in a hazelnut 2, homogenized samples are analyzed in the laboratory or measurement laboratory in the inventive method using available analytical laboratory methods, such as gas chromatography or FTIR, and based on a statistically large quantity, to determine their chemical content. The determined concentration is then correlated with the corresponding hyperspectral information. The detection photosensor 4' can then be calibrated using the correlation determined between the spectral response, such as wavelength and / or amplitude, and the substance concentrations. In the case of the rotten hazelnut 2, these are changes in the free fatty acids. It has also been shown that the chemical changes inside, for example, a hazelnut 2 correlate with the measurement at the surface.This also applies to oilseeds, nuts, or seeds. The inventive method is particularly effective for oilseeds, nuts, or seeds in which the seed coat is located in the optical path.

[0082] The solution according to the invention consists of searching for fatty acid degradation products in the spectral range used. The model is preferably not limited to a single chemical substance, but rather to five main components that exhibit the greatest concentration difference between good and spoiled oilseeds, nuts, especially hazelnuts, or seeds. This means that, in the relevant spectral range, the quantitative degree of flavor-altering substances can be automatically inferred from the measured values ​​of several substances and their absolute amplitudes in the spectrum. According to the invention, the measured value is then defined as the degree of spoilage, for example, in a range of 0-100%. Subsequently, sorting can be carried out based on the first and second threshold values, for example, in a sorting plant.

[0083] Preferably, the assignment of the at least one characteristic wavelength or at least one characteristic wavelength range of the recorded absorption or reflection spectrum of the sample to the degree of putrefaction is carried out within the framework of the inventive method by means of at least one average value, a bandwidth, or individual frequency bands of the recorded absorption or reflection spectra. This allows wavelengths or wavelength ranges representative of specific fatty acid degradation products to be used in the absorption or reflection spectrum to determine the degree of putrefaction. Furthermore, the acquisition of the absorption or reflection spectra by the calibration photosensor 4 and / or the detection photosensor 4' is preferably carried out by means of hyperspectral acquisition.Consequently, the calibration photosensor 4 and / or the detection photosensor 4' of the device 1 according to the invention are preferably designed as one hyperspectral camera or as two separate hyperspectral cameras. Hyperspectral cameras enable the use of the device 1 according to the invention, or the method according to the invention, in modern sorting systems in the food industry. Unlike normal color cameras, these capture not only the visible light spectrum but also additional spectral ranges, e.g., in the infrared range. Since the infrared images provide information about the chemical properties on the surface of the products using chemical imaging technology, they are ideally suited to depicting the quality of food products and detecting deficiencies, defects, or contaminations that are hidden from the human eye. In contrast to analytical methods in the laboratory, orIn measurement laboratories, these photographic analysis methods are suitable for checking high material throughputs of food with the help of automation technology and for separating them in sorting systems at high speed in real time.

[0084] To determine the concentration of at least one fatty acid degradation product in the sample, gas chromatography followed by mass spectrometric analysis is preferably performed. This allows for a precise determination of the fatty acid degradation product concentration in the sample.

[0085] As in Figure 1 and in Figure 2As shown, the device 1 according to the invention preferably comprises a sorting unit 5. This unit is configured to separate an oilseed, a nut, in particular a hazelnut 2, or a seed from a product stream when the oilseed, the nut, in particular the hazelnut 2, or the seed is classified as rotten. For this purpose, the sorting unit 5 can, for example, comprise a pneumatic deflection unit 6, which is connected to the detection photosensor 4'. The pneumatic deflection unit separates oilseeds, nuts, in particular hazelnuts 2, or seeds classified as rotten from a substantially continuous product stream.

[0086] The method according to the invention comprises the sensory detection of hazelnut kernels, or more generally of oilseeds, nuts or seeds, detected in reflectance or transmittance, in a wavelength range of 300-2500 nm or a subrange thereof, preferably by means of a hyperspectral or multispectral camera. The spatially resolved spectral data thus obtained are evaluated by a computing unit 8, such as a PC, an embedded system with FPGA, CPU or GPU, in such a way that they are correlated in a training mode with measured values ​​from a reference laboratory for the determination of chemical components in, for example, hazelnuts, e.g., using GCC.The correlation obtained during training—that is, the spectral information that can be related to the information in the reference dataset—is then used in an operating mode characterized by the detection steps to assign a corresponding measurement value for the degree of laziness to newly acquired spectra. Thus, each pixel in the camera's field of view corresponds to a reference value based on calibration through the established correlation between spectral data and the measurement laboratory.

[0087] Any suitable statistical method from the field of multivariate regression analysis can be used to extract the relevant correlation information, for example, by correlating the variance in the spectral dataset with that in the laboratory measurement dataset. Such methods are well-known without loss of generality, e.g., Principal Component Regression (PCR), Partial Least Squares Regression (PLSR), Multiple Linear Regression (MLR), Multivariate Curve Resolution (MCR), soft independent modeling of class analogy (SIMCA), support vector regression (SVR), regression using artificial neural networks, decision trees, and / or random forests.

[0088] Based on the spatially resolved optical measurement of the surfaces of oilseeds, nuts (especially hazelnuts), or seeds obtained in this way, it is possible to determine, by defining the first and second thresholds, whether, for example, an inspected almond is rotten or belongs to the good product. The first threshold defines the number of rot pixels per object at which, for example, an almond or a hazelnut should be classified as rotten. The second threshold defines at which a pixel value in the camera's field of view is to be counted as rotten.

[0089] Figure 4 shows an exemplary evaluation of fatty acid degradation products in rotten hazelnuts 2 within the framework of a selection of those fatty acid degradation products with the greatest differences between a rotten hazelnut and a non-rotten hazelnut.

[0090] Preferably, the fatty acid degradation products determined in the process according to the invention are butyrolactone, diacetyl, 2-methylbutanal, 3-methylbutanal, acetylacetone, filbertone, and 2,3-butanediol. Specifically, putridness is attributed to butyrolactone, a cyclic ester of hydroxycarboxylic acids. In particular, the identification of the fatty acid degradation products of the sample can be carried out by mass spectroscopic detection, specifically by generating chromatograms at mass-to-charge ratios selected in the range between 20 and 300, preferably at least one mass-to-charge ratio of 86 for butyrolactone, diacetyl, 2-methylbutanal, 3-methylbutanal, and acetylacetone; 45 for 2,3-butanediol; and 69 for filbertone. The mass-to-charge ratio of acetic acid is typically 60.

[0091] The device 1 according to the invention, which preferably comprises a sorting unit 5, and the method according to the invention are characterized in particular by the following advantages: High processing speed and decision reliability are provided in the inline sorting process, due to the evaluation of the correlations between the concentration of at least two fatty acid degradation products and the spectral response. Furthermore, a high-quality offline concentration determination based on the current state of offline laboratory technology is achieved, which can be used inline in the sorting process. According to the invention, the sorting of putrefaction is not a binary quantity, such as putrefied or not putrefied, but an analogous quantity that is based on the amplitude at certain wavelengths or spectral responses.This can be traced back to the average amplitudes in a wavelength range and a spatial distribution in, for example, a hazelnut (2), through surface measurement. Based on the spectral rottenness degree detection, sorting with high product quality requirements and precise setting of the sorting limit in the sorting system can be carried out. Reference symbol list

[0092] 2 Oilseed, especially hazelnut 3 Calibration light source 3 Detection light source 4 Calibration photosensor 4 Detection photosensor 5 Sorting unit 6 Deflection unit 7 Control unit 8 Computing unit 9 AI module

Claims

1. A method for determining whether an oleaginous fruit, a nut, in particular a hazelnut (2), or a seed is putrid, comprising at least the following calibration steps, preferably in the following order: a) irradiating a sample of an oleaginous fruit, a nut, in particular a hazelnut (2), or a seed with a calibration light source (3), b) capturing an absorption or reflection spectrum in a wavelength range of 300-2500 nm by means of a calibration photosensor (4), c) associating at least one characteristic wavelength or at least one characteristic wavelength range of the captured absorption or reflection spectrum of the sample with a degree of putrescence, d) repeating the preceding steps for a representative number of samples and forming a correlation between the degree of putrescence and the associated wavelengths or wavelength ranges of the captured absorption or reflection spectra, wherein the method then comprises the following detection steps, preferably in the following order: e) irradiating an oleaginous fruit, a nut, in particular a hazelnut (2), or a seed with a detection light source (3'), f) capturing absorption or reflection spectra in a wavelength range of 300-2500 nm by means of a detection photosensor (4'), wherein the detection photosensor comprises a plurality of pixels and captures one absorption or reflection spectrum per pixel in each case, g) associating a degree of putrescence with each pixel of the detection photosensor (4') by applying the correlation in accordance with step d) to the absorption or reflection spectra captured from the pixels of the detection photosensor (4'), h) classifying the oleaginous fruit, the nut, in particular the hazelnut (2), or the seed as putrid when at least a specific number of pixels has a degree of putrescence which exceeds a first threshold value and / or when a degree of putrescence associated with at least one pixel exceeds a second threshold value.

2. The method according to any of claims 1, characterized in that, before step b), the light reflected and / or transmitted by the sample is projected onto a calibration photosensor (4).

3. The method according to any of claims 1 or 2, characterized in that, after step b), at least one degree of putrescence is determined on the basis of the content of at least one fatty acid decomposition product in the sample, and in particular additionally on the basis of the content of an acetic acid in the sample.

4. The method according to claim 3, characterized in that the at least one fatty acid decomposition product comprises at least one component from the group of butyrolactone, diacetyl, 2-methylbutanal, 3-methylbutanal, acetylacetone, filbertone, or 2,3-butanediol and / or the degree of putrescence is determined on the basis of the content of at least two fatty acid decomposition products in the oleaginous fruit.

5. The method according to any of claims 3 or 4, characterized in that a plurality of components of the fatty acid decomposition products which have the greatest difference in concentration between good and putrid oleaginous fruits, nuts, in particular hazelnuts (2), or seeds are sought.

6. The method according to any of the preceding claims, characterized in that the correlation is at least one correlation function or at least one index table or a value table, or comprises at least one comparison of spectral information which comprises at least the degree of putrescence of the oleaginous fruit and the wavelengths or wavelength ranges of the captured absorption or reflection spectra associated therewith.

7. The method according to any of the preceding claims, characterized in that a shared sensor is used as the calibration photosensor (4) and as the detection photosensor (4') and / or one light source is used as the calibration light source (3) and as the detection light source (3').

8. The method according to any of the preceding claims, characterized in that the at least one characteristic wavelength or the at least one characteristic wavelength range of the captured absorption or reflection spectrum of the sample is associated with the degree of putrescence by means of at least one of an average, a bandwidth, or individual frequency bands of the captured absorption or reflection spectrum.

9. The method according to any of the preceding claims, characterized in that the absorption or reflection spectra are captured by the calibration photosensor (4) and / or the detection photosensor (4') by means of hyperspectral capturing and in particular by means of a color camera.

10. The method according to any of the preceding claims, characterized in that the correlation in an operating mode which is characterized by the detection steps is used to associate a corresponding measured value for the degree of putrescence with newly recorded spectra.

11. The method according to claim 10, characterized in that a reference value based on the calibration by means of the determined correlation between spectral data and a measuring laboratory corresponds to each pixel.

12. A device (1) for determining whether an oleaginous fruit, a nut, in particular a hazelnut (2), or a seed is putrid, comprising a calibration light source (3), a detection light source (3'), a calibration photosensor (4), a detection photosensor (4'), and an arithmetic logic unit (8), characterized in that the device (1) is configured to carry out a method according to any of claims 1 to 11.

13. The device (1) according to claim 12, characterized in that the device (1) comprises a sorting unit (5), and the sorting unit (5) in particular operates at least with two adjustable threshold values, a first adjustable threshold value comprising a substance quantity for a pixel and the second adjustable threshold value comprising an inhomogeneous distribution of the substance, by the second adjustable threshold value taking into consideration the number of pixels having the attribute: putrid for the sorting.

14. A computer program product comprising program commands which, when executed, are configured to cause a device according to claim 12 to carry out at least a method according to any of claims 1 to 11.

15. A computer-readable medium, which comprises at least one computer program product which, when executed by at least the arithmetic logic unit (8) of a device according to claim 12, causes said device to carry out at least a method according to any of claims 1 to 11.