A method and system for detecting tobacco leaf processing quality based on hyperspectral technology

Abstract

The present invention provides a method and system for detecting the quality of tobacco leaf processing based on hyperspectral technology, and relates to the technical field of tobacco leaf processing detection. The method comprises: fixing a tobacco leaf sample on a transparent module of a conveyor belt; moving the transparent module to a detection position; turning off an LED fill light, a halogen tungsten lamp and an ultraviolet light source, turning on a transmission light box, and obtaining transmission spectrum data of the tobacco leaf sample; fixing the tobacco leaf sample on a black module of a conveyor belt; moving the black module to a detection position; turning off the transmission light box and turning on an LED fill light, turning on a halogen tungsten lamp and an ultraviolet light source in sequence, and obtaining reflection spectrum data and fluorescence spectrum data of the tobacco leaf sample; and determining the comprehensive quality information of tobacco leaf processing according to the transmission spectrum data, reflection spectrum data and fluorescence spectrum data. According to the present invention, the irradiation of multiple light sources and the collection of multimodal hyperspectral data can be realized, the multiple quality characteristics of tobacco leaf samples can be evaluated, and the comprehensiveness, accuracy and flexibility of quality detection can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of tobacco leaf processing detection, and in particular to a tobacco leaf processing quality detection method and system based on hyperspectral technology. Background Art

[0002] In the related art, CN118730960A discloses an online multiple inspection method for the quality of reconstituted tobacco raw materials and its application, the method comprising the following steps: (1) collecting near-infrared spectral data of the sample to be tested, and analyzing to obtain the content of chemical indicators in the sample to be tested; (2) screening out key chemical indicators according to the content of chemical indicators, and establishing comprehensive indicators for quality evaluation; (3) for the test to be decided, assigning the significance level corresponding to the current test according to the adaptive algorithm, implementing the test, and determining whether the quality of the sample to be tested meets the standard; (4) calculating the false discovery rate and / or the false discovery frequency, and evaluating the test effect. The inspection method provided by the scheme can quickly and effectively realize online inspection, which can not only overcome the deficiency of excessively large cumulative significance level, but also avoid the risk of too small significance level at the end of the hypothesis test sequence. Compared with the existing offline inspection method, the online multiple inspection method provided by the scheme can be carried out continuously for a long time.

[0003] CN118583790A discloses a method, medium and system for rapid analysis of the spectrum of baking characteristics of fresh tobacco leaves, belonging to the technical field of tobacco leaf baking. The method for rapid analysis of the spectrum of baking characteristics of fresh tobacco leaves comprises the following steps: obtaining representative spectral information of fresh tobacco leaf samples harvested from the field, recorded as tobacco leaf spectrum; obtaining the baking characteristics of the fresh tobacco leaf samples in a dark box experiment, evaluating and grading them to obtain baking characteristic labels; obtaining the average spectrum of the same fresh tobacco leaf, establishing a mapping relationship between the average spectrum of the fresh tobacco leaf and the baking characteristic label, recorded as spectral label mapping, thereby obtaining a fresh tobacco leaf spectrum data set with different baking characteristic labels; training a neural network with the fresh tobacco leaf spectrum data set to obtain a fresh tobacco leaf baking characteristic discrimination model; obtaining a test image of the fresh tobacco leaf to be tested; obtaining the test spectrum of the fresh tobacco leaf according to the test image; discriminating the test spectrum using the discrimination model to obtain a discrimination result; the scheme can analyze the baking characteristics of fresh tobacco leaves through spectral detection technology.

[0004] Therefore, although the relevant technology can detect the quality and characteristics of tobacco leaves, the relevant technology mainly uses a single spectrum for detection. However, the spectral data generated by different light sources can be used to detect different quality characteristics, and the relevant technology cannot flexibly switch light sources, nor can it use multiple spectral data to comprehensively detect tobacco leaves. Therefore, it is impossible to obtain more comprehensive quality information.

[0005] The information disclosed in the background technology section of this application is only intended to deepen the understanding of the general background technology of this application, and should not be regarded as an admission or any form of suggestion that the information constitutes the prior art already known to those skilled in the art. Summary of the invention

[0006] The present invention provides a tobacco leaf processing quality detection method and system based on hyperspectral technology, which can solve the technical problems that related technologies are difficult to flexibly switch light sources and difficult to obtain relatively comprehensive quality information.

[0007] According to a first aspect of the present invention, a method for detecting tobacco processing quality based on hyperspectral technology is provided, comprising: fixing each tobacco leaf sample in a tobacco leaf sample group to be detected on a transparent module of a conveyor belt, wherein the conveyor belt comprises a plurality of alternately distributed transparent modules and black modules, and the sizes and quantities of the transparent modules and the black modules are the same, the transparent modules are carriers made of transparent flexible materials, the black modules are carriers made of black opaque flexible materials, and a plurality of alternately distributed transparent modules and black modules are interconnected to form the conveyor belt; controlling the movement of the conveyor belt, and moving each transparent module to a detection position, wherein the detection position comprises a transmission light box, an LED fill light, a halogen tungsten lamp, an ultraviolet light source and a hyperspectral imager, the transmission light box is located directly below the conveyor belt, and is used to emit a narrow band transmission light beam directly above, the LED fill light is located obliquely above the conveyor belt, and is used to fill light for the tobacco leaf sample, the halogen tungsten lamp is located on an annular lamp stand, the annular lamp stand is located directly above the conveyor belt, the halogen tungsten lamp is used to irradiate the tobacco leaf sample with light of a preset wavelength band, and the ultraviolet light source is located obliquely above the conveyor belt. The transmission light box is kept closed, and the halogen tungsten lamp and the ultraviolet light source are kept turned on, and when each transparent module moves to the detection position, the transmission spectrum data of the tobacco leaf sample is obtained through the hyperspectral imager; after obtaining the transmission spectrum data of each tobacco leaf sample in the tobacco leaf sample group to be detected, each tobacco leaf sample is fixed on the black module of the conveyor belt respectively; the conveyor belt is controlled to move, and each black module is moved to the detection position; the transmission light box is kept closed and the LED fill light is kept turned on, and when each black module moves to the detection position, the halogen tungsten lamp and the ultraviolet light source are turned on in sequence; when the halogen tungsten lamp is turned on, the reflection spectrum data of the tobacco leaf sample is obtained through the hyperspectral imager, and when the ultraviolet light source is turned on, the fluorescence spectrum data of the tobacco leaf sample is obtained through the hyperspectral imager; according to the transmission spectrum data, reflection spectrum data and fluorescence spectrum data of multiple tobacco leaf samples, the comprehensive quality information of tobacco leaf processing is determined.

[0008] According to a second aspect of the present invention, a tobacco processing quality detection system based on hyperspectral technology is provided, comprising: a first fixing module, used to fix each tobacco leaf sample in the tobacco leaf sample group to be detected on a transparent module of a conveyor belt, wherein the conveyor belt comprises a plurality of alternately distributed transparent modules and black modules, and the size and quantity of the transparent modules and the black modules are the same, the transparent module is a carrier made of a transparent flexible material, the black module is a carrier made of a black opaque flexible material, and a plurality of alternately distributed transparent modules and black modules are interconnected to form the conveyor belt; a first moving module, used to The conveyor belt is controlled to move, and each transparent module is moved to a detection position, wherein the detection position includes a transmission light box, an LED fill light, a halogen tungsten lamp, an ultraviolet light source and a hyperspectral imager, wherein the transmission light box is located directly below the conveyor belt, and is used to emit a narrow band transmission light beam directly upward, the LED fill light is located obliquely above the conveyor belt, and is used to fill light for tobacco leaf samples, the halogen tungsten lamp is located on an annular lamp stand, and the annular lamp stand is located directly above the conveyor belt, the halogen tungsten lamp is used to irradiate tobacco leaf samples with light of a preset wavelength band, the ultraviolet light source is located obliquely above the conveyor belt, and is used to emit ultraviolet light to irradiate tobacco leaf samples, and the The hyperspectral imager is located directly above the annular light stand. The transmitted light beam is received by the hyperspectral imager after passing through the transparent module, the tobacco leaf sample and the annular light stand. The first acquisition module is used to keep the LED fill light, the halogen tungsten lamp and the ultraviolet light source turned off, keep the transmission light box turned on, and obtain the transmission spectrum data of the tobacco leaf sample through the hyperspectral imager when each transparent module moves to the detection position. The second fixing module is used to fix each tobacco leaf sample on the black module of the conveyor belt after obtaining the transmission spectrum data of each tobacco leaf sample in the tobacco leaf sample group to be detected. The second motion module is used to control the conveyor belt. Movement, and move each black module to the detection position; a light source control module, used to keep the transmission light box closed and keep the LED fill light turned on, and when each black module moves to the detection position, turn on the halogen tungsten lamp and the ultraviolet light source in sequence; a second acquisition module, used to obtain the reflection spectrum data of the tobacco leaf sample through the hyperspectral imager when the halogen tungsten lamp is turned on, and to obtain the fluorescence spectrum data of the tobacco leaf sample through the hyperspectral imager when the ultraviolet light source is turned on; a processing comprehensive quality information module, used to determine the comprehensive quality information of tobacco leaf processing according to the transmission spectrum data, reflection spectrum data and fluorescence spectrum data of multiple tobacco leaf samples.

[0009] Technical effect: According to the present invention, a specific transmission belt can be used to carry tobacco leaf samples, so that transparent modules and black modules are used to provide different light sources with the environment required for detecting hyperspectral data, thereby improving the detection accuracy of hyperspectral data. Different types of hyperspectral data can be detected by controlling different light sources to illuminate tobacco leaf samples. Without changing the transmission belt and light source, multiple light sources can be used for irradiation and multimodal hyperspectral data collection and analysis, thereby determining multiple quality characteristics of tobacco leaf samples and improving the comprehensiveness, accuracy and flexibility of quality detection. When determining the quality score of conventional chemical components, the minimum quality level of the batch of tobacco leaf samples in terms of conventional chemical components can be described by the maximum value of the relative difference between the sampling transmission spectrum function and the quality standard transmission spectrum function, which is used to represent the overall quality level of the batch of tobacco leaf samples in terms of conventional chemical components, thereby representing the overall closeness of the batch of tobacco leaf samples in terms of conventional chemical components to the standard conventional chemical components, and obtaining the quality score of conventional chemical components, which can objectively and accurately describe the quality characteristics of the batch of tobacco leaf samples in terms of conventional chemical components. When determining the quality score of surface solid non-tobacco debris, the conditional function can be used to find the minimum and maximum points in the reflection spectrum data, and record the corresponding wavelengths, and then determine whether the reflection spectrum sampling data has the characteristics of the reflection peak and absorption peak of specific solid non-tobacco debris by cosine similarity, so as to determine whether there is specific solid non-tobacco debris on the surface of the tobacco leaf sample, and improve the accuracy of solid non-tobacco debris detection. In addition, the quality of the tobacco leaf sample with the most serious residual solid non-tobacco debris can be used to represent the overall quality of the batch of tobacco leaves in terms of surface solid non-tobacco debris, so as to improve the accuracy and objectivity of the quality score of solid non-tobacco debris. When determining the quality score of the degree of mold, the fluorescence spectrum wavelength of each type of moldy product can be substituted into the sampled fluorescence spectrum function to determine the light intensity of the fluorescence spectrum function at this wavelength, so as to determine whether the moldy product exceeds the standard, and then set the quality function of this type of moldy product through the conditional function. Furthermore, the quality status of tobacco leaf samples in terms of mildew can be solved by weighted summation, and then the minimum value of the corresponding quality status of each tobacco leaf sample in terms of mildew can be taken to describe the overall quality status of the batch of tobacco leaf samples in terms of the control of the content of mildew products, thereby improving the accuracy and objectivity of the mildew degree quality score.

[0010] It should be understood that the above general description and the following detailed description are exemplary and explanatory only and do not limit the present invention. Other features and aspects of the present invention will become more apparent from the following detailed description of exemplary embodiments with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, other embodiments can be obtained based on these drawings without creative work.

[0012] Figure 1 A flowchart of a tobacco leaf processing quality detection method based on hyperspectral technology according to an embodiment of the present invention is exemplarily shown;

[0013] Figure 2 A schematic diagram exemplarily shows a conveyor belt according to an embodiment of the present invention;

[0014] Figure 3 A schematic diagram exemplarily showing a detection position according to an embodiment of the present invention;

[0015] Figure 4 A block diagram of a tobacco leaf processing quality detection system based on hyperspectral technology according to an embodiment of the present invention is exemplarily shown. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] The technical solution of the present invention is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0018] Figure 1A flowchart of a tobacco processing quality detection method based on hyperspectral technology according to an embodiment of the present invention is exemplarily shown, the method comprising: step S101, fixing each tobacco leaf sample in a tobacco leaf sample group to be detected on a transparent module of a conveyor belt, wherein the conveyor belt comprises a plurality of alternately distributed transparent modules and black modules, and the size and quantity of the transparent modules and the black modules are the same, the transparent module is a carrier made of a transparent flexible material, the black module is a carrier made of a black opaque flexible material, and a plurality of alternately distributed transparent modules and black modules are interconnected to form the conveyor belt; Step S102, controlling the movement of the conveyor belt and moving each transparent module to a detection position, wherein the detection position includes a transmission light box, an LED fill light, a halogen tungsten lamp, an ultraviolet light source and a hyperspectral imager, wherein the transmission light box is located directly below the conveyor belt and is used to emit a narrow band transmission light beam directly above, the LED fill light is located obliquely above the conveyor belt and is used to fill light for the tobacco leaf sample, the halogen tungsten lamp is located on an annular lamp stand, and the annular lamp stand is located directly above the conveyor belt, the halogen tungsten lamp is used to irradiate the tobacco leaf sample with light of a preset wavelength band, and the ultraviolet light source is located obliquely above the conveyor belt and is used to emit ultraviolet light. The hyperspectral imager is located directly above the annular light stand, and the transmitted light beam is received by the hyperspectral imager after passing through the transparent module, the tobacco leaf sample and the annular light stand; step S103, keeping the LED fill light, the halogen tungsten lamp and the ultraviolet light source turned off, keeping the transmission light box turned on, and when each transparent module moves to the detection position, the transmission spectrum data of the tobacco leaf sample is obtained through the hyperspectral imager; step S104, after obtaining the transmission spectrum data of each tobacco leaf sample in the tobacco leaf sample group to be detected, each tobacco leaf sample is fixed on the black module of the conveyor belt; step S10 5. Control the movement of the conveyor belt and move each black module to the detection position; step S106, keep the transmission light box closed and keep the LED fill light turned on, and when each black module moves to the detection position, turn on the halogen tungsten lamp and the ultraviolet light source in sequence; step S107, when the halogen tungsten lamp is turned on, obtain the reflection spectrum data of the tobacco leaf sample through the hyperspectral imager, and when the ultraviolet light source is turned on, obtain the fluorescence spectrum data of the tobacco leaf sample through the hyperspectral imager; step S108, determine the comprehensive quality information of tobacco leaf processing according to the transmission spectrum data, reflection spectrum data and fluorescence spectrum data of multiple tobacco leaf samples.

[0019] According to the tobacco processing quality detection method based on hyperspectral technology of the embodiment of the present invention, a specific transmission belt can be used to carry tobacco samples, so that the transparent module and the black module can be used to provide different light sources with the environment required for detecting hyperspectral data, thereby improving the detection accuracy of hyperspectral data. Different types of hyperspectral data can be detected by controlling different light sources to illuminate tobacco samples, and multiple light sources can be used for illumination and multi-modal hyperspectral data collection and analysis without replacing the transmission belt and light source, thereby determining multiple quality characteristics of tobacco samples and improving the comprehensiveness, accuracy and flexibility of quality detection.

[0020] According to one embodiment of the present invention, in step S101, the tobacco leaf sample group to be tested may include multiple tobacco leaf samples, and the multiple tobacco leaf samples may come from the same batch of tobacco leaf processing products. For example, the same batch of tobacco leaf processing products may be sampled to obtain multiple tobacco leaf samples to constitute the tobacco leaf sample group to be tested.

[0021] Figure 2 A schematic diagram of a conveyor belt according to an embodiment of the present invention is exemplarily shown.

[0022] According to one embodiment of the present invention, Figure 2 The top view of the conveyor belt. Figure 2 The white rectangular block in the middle is a transparent module, and the shadowed rectangular frame is a black module. The transparent modules and the black modules are distributed alternately, and on the entire conveyor belt, the number of transparent modules and black modules is the same, and the size of each transparent module and each black module is the same. The transparent module is a carrier made of a transparent flexible material, for example, a carrier made of a transparent PVC material, which can be used to carry tobacco leaf samples, and after the transmission light source on one side of the transparent module emits light to it, the light can be projected from the transparent module and the tobacco leaf sample, and the projected light can be received by the hyperspectral imager on the other side of the transparent module, and the transmission spectrum data of the tobacco leaf sample can be detected. The black module is a carrier made of a black opaque flexible material, for example, a carrier made of a black opaque PU material, and the color of the connection between the transparent module and the black module is also black. When a black module is used to carry tobacco leaf samples, a reflective light source or a fluorescent light source can be used to illuminate the tobacco leaf samples to make the tobacco leaf samples reflect light or fluoresce. Moreover, due to the light absorption characteristics of the black module, it itself does not produce reflection or fluorescence, so that the reflection or fluorescence produced by the tobacco leaf samples is less interfered by the light in the environment, thereby improving the accuracy of the reflection spectrum data and the fluorescence spectrum data. The reflection or fluorescence produced by the tobacco leaf samples can be received by a hyperspectral imager located on the same side of the reflective light source and the fluorescent light source to detect the reflection spectrum data and the fluorescence spectrum data.

[0023] According to one embodiment of the present invention, in step S102, when detecting the transmission spectral data of the tobacco leaf sample, the driving wheel can be controlled to rotate by a motor to drive the conveyor belt to move. When detecting the transmission spectral data of the tobacco leaf sample, the conveyor belt can be controlled to move the transparent module to the detection position to detect the transmission spectral data of the tobacco leaf sample through the transmission light source and the hyperspectral imager.

[0024] Figure 3 A schematic diagram exemplarily shows a detection position according to an embodiment of the present invention.

[0025] According to one embodiment of the present invention, Figure 3 It is used to show the relative positions between multiple devices set at the detection position, and there is no restriction on the shape and size of each device, as well as the specific position data between the devices. Among them, the transmission light box is located directly below the conveyor belt as the above-mentioned transmission light source, and is used to emit a narrow band transmission light beam directly above, and the length direction of the narrow band is consistent with the width direction of the conveyor belt. After penetrating the transparent module and the tobacco leaf sample, the transmission light beam passes through the annular hollow position of the circular light stand and is received by the hyperspectral imager, so that the transmission spectrum data of the tobacco leaf sample can be analyzed. The halogen tungsten lamp is located on the circular light stand and can be used as a reflective light source. The ultraviolet light source is located obliquely above the conveyor belt and can be used as a fluorescent light source. When the black module carries the tobacco leaf sample and moves to the detection position, the halogen tungsten lamp can illuminate the tobacco leaf sample, and the tobacco leaf sample can form reflected light. The reflected light can pass through the annular hollow position of the circular light stand and be received by the hyperspectral imager, so as to analyze the reflected spectrum data of the tobacco leaf sample. The ultraviolet light source can emit ultraviolet light to illuminate the tobacco leaf samples, so that the tobacco leaf samples are stimulated to produce fluorescence, and are received by the hyperspectral imager through the circular hollow position of the circular light stand, so as to analyze the fluorescence spectrum data of the tobacco leaf samples. The LED fill light can be kept off when detecting the transmission spectrum data to avoid interference with the transmitted light. The LED fill light can be kept on when detecting the reflection spectrum data and the fluorescence spectrum data, so as to fill the visible light band. The hyperspectral data collection environment of the tobacco leaf samples at the detection position is a black closed dark box environment, that is, the transmission light box, LED fill light, ultraviolet light source, halogen tungsten lamp, hyperspectral imaging instrument and tobacco leaves to be detected. The main elements of the "light source-sample-receiver" data collection process all work in a black closed dark box environment; the tobacco leaf samples at the non-detection position are in a semi-open environment.

[0026] According to one embodiment of the present invention, in step S103, when the transparent module moves to the detection position, the LED fill light, the halogen tungsten lamp and the ultraviolet light source can be kept off, the transmission light box can be kept on, and the transmission spectrum data of the tobacco leaf sample can be obtained through the hyperspectral imager. In addition, after the transmission spectrum data of a tobacco leaf sample is obtained, the conveyor belt can be controlled to move, the next transparent module can be moved to the detection position, and the transmission spectrum data of the next tobacco leaf sample can be obtained. This step can be iteratively performed until the transmission spectrum data of the tobacco leaf samples on all transparent modules on the conveyor belt are obtained.

[0027] According to one embodiment of the present invention, in step S104, since the number of black modules is consistent with that of transparent modules, each tobacco leaf sample on the transparent module can be moved to the black module and fixed, so that when the black module is moved to the detection position, the reflection spectrum data and fluorescence spectrum data of the tobacco leaf samples on the black module can be obtained.

[0028] According to one embodiment of the present invention, in step S105, the conveyor belt can be controlled to move and the black module can be moved to the detection position to be irradiated by the reflective light source and the fluorescent light source respectively, so as to obtain the reflective spectrum data and the fluorescent spectrum data of the tobacco leaf sample respectively through the hyperspectral imager.

[0029] According to one embodiment of the present invention, in step S106, since the black module is opaque, the transmission light box has no effect when irradiating the black module, so the transmission light box can be kept closed to save electricity. When detecting the reflected spectrum data, the wavelength range of the light of the halogen tungsten lamp is usually between 400nm-2000nm, and the wavelength range of the light of the LED fill light is between 340nm-800nm. The LED fill light can be used to supplement the brightness of the visible light band to improve the light source coverage intensity of the red light and near-infrared part. When detecting the fluorescent spectrum data, the wavelength range of the ultraviolet light source is between 320nm-400nm, and the LED fill light can also supplement the light intensity of the ultraviolet band. Therefore, the LED fill light can be kept turned on, and the halogen tungsten lamp and the ultraviolet light source can be turned on in sequence. When the halogen tungsten lamp is turned on, the ultraviolet light source is turned off, the tobacco leaf sample reflects the light of the halogen tungsten lamp, and its background (i.e., the black module) can absorb the halogen tungsten lamp light, and the reflection intensity of the light is small, thereby reducing the influence of the background reflected light on the reflected spectrum data of the tobacco leaf sample. When the ultraviolet light source is turned on, the halogen tungsten lamp is turned off, and the tobacco leaf sample is excited by the ultraviolet light source to produce fluorescence. The background (i.e., the black module) can absorb the ultraviolet light and almost no reflection and fluorescence are produced, thereby reducing the impact of the background on the fluorescence spectrum data of the tobacco leaf sample.

[0030] According to one embodiment of the present invention, in step S107, the reflection spectrum data of the tobacco leaf sample is obtained by the hyperspectral imager when the halogen tungsten lamp is turned on, and the fluorescence spectrum data of the tobacco leaf sample is obtained by the hyperspectral imager when the ultraviolet light source is turned on. After the reflection spectrum data and fluorescence spectrum data of a tobacco leaf sample are detected, the next black module can be moved to the detection position, and the reflection spectrum data and fluorescence spectrum data of the tobacco leaf sample carried on the next black module can be obtained. The above process can be iteratively performed until the reflection spectrum data and fluorescence spectrum data of the tobacco leaf samples on all black modules are obtained.

[0031] According to one embodiment of the present invention, the above process can obtain three spectral data of each tobacco leaf sample, so that the quality characteristics of multiple dimensions can be detected based on the three spectral data, which can improve the comprehensiveness of the detection. In addition, during the detection process, there is no need to replace the conveyor belt or the detection equipment. It is only necessary to flexibly control the movement of the conveyor belt and the switching of the light source to obtain the three spectral data, which improves the flexibility of the detection. Further, during the detection process, a suitable background is provided for different light sources to reduce the interference of the background on the spectral data. For example, for the detection of transmission spectral data, a transparent module can be provided as the background, which does not block the transmission light and reduces the interference with the transmission spectral data. For the detection of reflection spectral data and fluorescence spectral data, a black module can be provided as the background. The background hardly reflects light and hardly produces fluorescence, thereby reducing the interference with the reflection spectral data and the fluorescence spectral data and improving the detection accuracy.

[0032] According to one embodiment of the present invention, in step S108, multi-dimensional quality analysis can be performed using the transmission spectrum data, reflection spectrum data and fluorescence spectrum data of the multiple tobacco leaf samples obtained above, thereby determining the comprehensive processing quality information of the tobacco leaf sample group to be tested.

[0033] According to one embodiment of the present invention, the comprehensive processing quality information of the tobacco leaf sample group to be tested is determined based on the transmission spectral data, reflection spectral data and fluorescence spectral data of multiple tobacco leaf samples, including: obtaining the conventional chemical component quality score of the tobacco leaf sample group to be tested based on the transmission spectral data of multiple tobacco leaf samples; obtaining the solid non-tobacco impurity quality score of the tobacco leaf sample group to be tested based on the reflection spectral data of multiple tobacco leaf samples; obtaining the mold degree quality score of the tobacco leaf sample group to be tested based on the fluorescence spectral data of multiple tobacco leaf samples; obtaining the processing quality score of the tobacco leaf sample group to be tested based on the conventional chemical component quality score, the solid non-tobacco impurity quality score and the mold degree quality score; and determining the comprehensive processing quality information based on the processing quality score of the tobacco leaf sample group to be tested.

[0034] According to one embodiment of the present invention, the transmitted light can penetrate the tobacco leaf sample, thereby reflecting the internal quality safety of the tobacco leaf sample, for example, whether the internal conventional chemical components meet the quality standards. According to the transmission spectrum data of multiple tobacco leaf samples, the conventional chemical component quality score of the tobacco leaf sample group to be tested is obtained, including: sampling the transmission spectrum data of each tobacco leaf sample to obtain transmission spectrum sampling data; fitting the transmission spectrum sampling data to obtain a sampled transmission spectrum function; according to the preset quality grade of the tobacco leaf sample group to be tested, determining the standard transmission spectrum data corresponding to the preset quality grade; fitting the standard transmission spectrum data to obtain the quality standard transmission spectrum function; according to the sampled transmission spectrum function and the quality standard transmission spectrum function, the conventional chemical component quality score is obtained.

[0035] According to one embodiment of the present invention, when sampling the transmission spectrum data of a tobacco leaf sample, uniform sampling can be performed to obtain transmission spectrum sampling data, and the difference between the transmission spectrum sampling data and the standard transmission spectrum data corresponding to the preset quality grade can be compared to determine whether the tobacco leaf sample meets the quality standard. For example, the tobacco leaf sample group to be tested has a preset quality grade, and the quality grade has corresponding standard transmission spectrum data. The standard transmission spectrum data can be obtained by performing transmission spectrum analysis on tobacco leaves that have been identified to meet the preset quality grade, or can be generated by a neural network model based on the content of various conventional chemical components in tobacco leaves that meet the quality grade. The present invention does not limit the method for obtaining the standard transmission spectrum data. The transmission spectrum sampling data can be fitted to obtain a sampled transmission spectrum function. For example, a polynomial fitting method can be used to fit the data to obtain a sampled transmission spectrum function. The standard transmission spectrum data can also be fitted to obtain a quality standard transmission spectrum function. When fitting, the same method as the transmission spectrum sampling data can be used to fit the data so that the two are comparable.

[0036] According to one embodiment of the present invention, obtaining a conventional chemical component quality score according to the sample transmission spectrum function and the quality standard transmission spectrum function includes: obtaining a conventional chemical component quality score according to formula (1): ,

[0037] (1),

[0038] in, is the sampling transmission spectrum function of the i-th tobacco leaf sample, is the mass standard transmission spectrum function, is the starting wavelength of the mass standard transmission spectrum function, is the end wavelength of the quality standard transmission spectrum function, n is the number of tobacco leaf samples, max is the maximum value function, is the wavelength variable, i≤n, and i and n are both positive integers.

[0039] According to one embodiment of the present invention, It is the integral value of the absolute value of the difference between the sampled transmission spectrum function of the i-th tobacco leaf sample and the quality standard transmission spectrum function, which can represent the total difference between the starting wavelength and the ending wavelength of the quality standard transmission spectrum function. The use of absolute values ​​can prevent the positive and negative differences from offsetting each other. It can represent the relative difference between the sampled transmission spectrum function of the ith tobacco leaf sample and the transmission spectrum function of the quality standard. By taking the maximum value of the relative difference corresponding to each tobacco leaf sample, the maximum value of the relative difference corresponding to the tobacco leaf samples of the batch can be determined, that is, the tobacco leaf sample in the batch of tobacco leaf samples whose conventional chemical components are least in line with the standards can be determined, that is, the worst quality situation can be determined, and then 1 minus the maximum value of the relative difference can be used as the quality score of the conventional chemical components, which can be used to describe the similarity between the conventional chemical components of the tobacco leaf samples of the batch and the conventional chemical components of the standards under the worst quality situation, and can also describe the overall quality level of the tobacco leaf samples of the batch in terms of conventional chemical components, that is, the higher the quality score of the conventional chemical components, the higher the overall quality level of the tobacco leaf samples of the batch in terms of conventional chemical components.

[0040] In this way, the minimum quality level of the batch of tobacco leaf samples in terms of conventional chemical components can be described by the maximum value of the relative difference between the sampling transmission spectrum function and the quality standard transmission spectrum function, which is used to represent the overall quality level of the batch of tobacco leaf samples in terms of conventional chemical components, thereby representing the overall degree of proximity of the batch of tobacco leaf samples to the standard conventional chemical components in terms of conventional chemical components, and obtaining the quality score of conventional chemical components, which can objectively and accurately describe the quality characteristics of the batch of tobacco leaf samples in terms of conventional chemical components.

[0041] According to one embodiment of the present invention, reflected light is light reflected from the surface of a tobacco leaf sample. Light of a certain wavelength band reflects light having specific spectral characteristics after being absorbed and reflected by substances on the surface of the tobacco leaf sample. For example, when certain solid non-tobacco debris (e.g., feathers, plastic particles, leather scraps, metal slag) exist on the surface, these non-tobacco substances may form reflected light having specific spectral characteristics after absorption and reflection. Therefore, the specific spectral characteristics may be used to determine whether certain solid non-tobacco debris are attached and mixed on the tobacco leaf surface.

[0042] According to one embodiment of the present invention, a solid non-tobacco impurity quality score of a tobacco leaf sample group to be tested is obtained based on reflectance spectrum data of multiple tobacco leaf samples, including: sampling the reflectance spectrum data of each tobacco leaf sample to obtain reflectance spectrum sampling data; setting a first wavelength corresponding to an absorption peak and a second wavelength corresponding to a reflection peak of multiple solid non-tobacco impurities; and determining the solid non-tobacco impurity quality score based on the reflectance spectrum sampling data, the first wavelength, and the second wavelength.

[0043] According to one embodiment of the present invention, the reflectance spectrum data of each tobacco leaf sample can be sampled to obtain the reflectance spectrum sampling data. The sampling method is similar to the sampling method of the transmission spectrum sampling data described above, and will not be repeated here. Specific solid non-tobacco debris has specific spectral characteristics. For example, it absorbs light of a certain wavelength and reflects light of a certain wavelength. Therefore, the first wavelength corresponding to its absorption peak and the second wavelength corresponding to its reflection peak can be determined. Then, the first wavelength and the second wavelength are used to compare with the reflectance spectrum sampling data to determine whether the reflectance spectrum sampling data has significant absorption and reflection effects at the first wavelength and the second wavelength, thereby determining whether the tobacco leaf sample surface is attached to and mixed with such solid non-tobacco debris.

[0044] According to an embodiment of the present invention, determining the solid non-smoking debris quality score according to the reflection spectrum sampling data, the first wavelength and the second wavelength includes: determining the solid non-smoking debris quality score according to formula (2): ,

[0045] (2),

[0046] in, is the relative illumination data of the reflection spectrum sampling data of the i-th tobacco leaf sample at the j-th sampling point, is the relative illumination data of the reflectance spectrum sampling data of the i-th tobacco leaf sample at the j+1-th sampling point, is the relative illumination data of the reflection spectrum sampling data of the i-th tobacco leaf sample at the j-1-th sampling point, m is the number of sampling points for sampling the reflection spectrum data, and is the logical operator of "and", for The transposed vector of is the first wavelength corresponding to the kth solid non-smoke debris, is the second wavelength corresponding to the kth solid non-smoking debris cigarette piece, is the wavelength corresponding to the jth sampling point, is the preset threshold, is the number of types of solid non-tobacco debris, n is the number of tobacco leaf samples, max is the maximum value function, if is the conditional function, i≤n, k≤ , 2≤j≤m-1, and i, n, k, , j and m are all positive integers.

[0047] According to one embodiment of the present invention, in formula (2), In the reflection spectrum sampling data of the i-th tobacco leaf sample, if the data of the j-th sampling point is less than or equal to the previous sampling data, and less than or equal to the next sampling data, that is, the j-th sampling point is a minimum point in the reflection spectrum data, it can be said that the light intensity at this point is absorbed by some substances, which may be the first wavelength corresponding to the absorption peak of some substances. Therefore, the wavelength corresponding to its serial number can be recorded. Otherwise, the conditional function value is set to 0. Therefore, if all sampling points are traversed, the serial numbers of all minimum points can be found, that is, the set of values ​​of the conditional function includes the wavelengths corresponding to the serial numbers of all minimum points and 0.

[0048] According to one embodiment of the present invention, in formula (2), In the reflection spectrum sampling data of the i-th tobacco leaf sample, if the data of the j-th sampling point is greater than or equal to the previous sampling data, and greater than or equal to the next sampling data, that is, the j-th sampling point is a maximum point in the reflection spectrum data, it can be indicated that there is reflected light of some substance at this wavelength, which may be the second wavelength corresponding to the reflection peak of some substance. Therefore, the wavelength corresponding to its serial number can be recorded, otherwise, the conditional function value is set to 0. Therefore, if all sampling points are traversed, the serial numbers of all maximum points can be found, that is, the set of values ​​of the conditional function includes the wavelengths corresponding to the serial numbers of all maximum points and 0.

[0049] According to one embodiment of the present invention, the vector is a vector consisting of one value selected from the set of values ​​of the above two conditional functions, Then it can represent the maximum value of the cosine similarity between the above vector and the vector composed of the first wavelength and the second wavelength corresponding to the k-th solid non-smoking debris. That is, one value can be selected from the set of values ​​of the above two conditional functions to form a vector, and its similarity with The cosine similarity between can be traversed through all values ​​in the set of values ​​of the above two conditional functions to find the maximum value of the cosine similarity. It can be said that if the maximum value of the above cosine similarity is greater than or equal to the preset threshold, it means There is a wavelength in the value set of that is close to or consistent with the first wavelength corresponding to the k-th solid non-smoking debris, and There is a wavelength close to or consistent with the second wavelength corresponding to the k-th solid non-smoking impurity in the value set of . Therefore, the spectral characteristics of the reflectance spectrum sampling data can reflect the absorption peak and reflection peak of the k-th solid non-smoking impurity. That is, if there is the k-th solid non-smoking impurity on the surface of the ith tobacco leaf sample, the conditional function value can be 1. Otherwise, it is considered that there is no k-th solid non-smoking impurity on the surface of the ith tobacco leaf sample, and the conditional function value can be 0. The number of types of solid non-tobacco debris on the surface of the i-th tobacco leaf sample can be counted, and the tobacco leaf sample with the largest number of types of solid non-tobacco debris on the surface can be selected. The ratio of the number of types of solid non-tobacco debris on its surface to the total number of solid non-tobacco debris types indicates the most serious pollution condition of the batch of tobacco leaf samples, which can be used to represent the overall pollution condition of the batch of tobacco leaf samples. Furthermore, the surface solid non-tobacco debris quality score can be obtained by subtracting the ratio from 1 to indicate the overall quality of the batch of tobacco leaves in terms of solid non-tobacco debris. The higher the surface solid non-tobacco debris quality score, the less solid non-tobacco debris residues in the batch of tobacco leaves, and the better the quality.

[0050] In this way, the minimum and maximum points in the reflectance spectrum data can be found through the conditional function, and the corresponding wavelengths can be recorded. Then, the cosine similarity method can be used to determine whether the reflectance spectrum sampling data has the characteristics of the reflection peak and absorption peak of specific solid non-tobacco debris, thereby determining whether there are specific solid non-tobacco debris on the surface of the tobacco leaf sample, and improving the accuracy of solid non-tobacco debris detection. In addition, the quality of the tobacco leaf samples with the most serious residual solid non-tobacco debris can be used to represent the overall quality of the batch of tobacco leaves in terms of solid non-tobacco debris, improving the accuracy and objectivity of the solid non-tobacco debris quality score.

[0051] According to one embodiment of the present invention, fluorescence spectral data can be used to determine whether specific mold products exist in tobacco leaf samples, as well as the content of specific mold products, such as mold products of Penicillium and Mucor fungi, and Bacillus, Lactobacillus and Pectobacillus bacteria. Tobacco leaves of good quality can control specific mold products at a certain level, and fluorescence spectral data can be used to detect whether the content of specific mold products is higher than a preset level.

[0052] According to one embodiment of the present invention, based on the fluorescence spectrum data of multiple tobacco leaf samples, a mold degree quality score of the tobacco leaf sample group to be tested is obtained, including: sampling the fluorescence spectrum data of each tobacco leaf sample to obtain fluorescence spectrum sampling data; fitting the fluorescence spectrum sampling data to obtain a sampled fluorescence spectrum function; determining the fluorescence spectrum wavelength of a preset moldy product; and obtaining the mold degree quality score based on the sampled fluorescence spectrum function and the fluorescence spectrum wavelength.

[0053] According to one embodiment of the present invention, the fluorescence spectrum data can be sampled, and after sampling, fitting can be performed to obtain a sampled fluorescence spectrum function. The sampling and fitting methods are similar to the sampling and fitting methods of the above-mentioned transmission spectrum sampling data, and will not be repeated here. The above-mentioned preset moldy product can have a specific fluorescence spectrum wavelength, that is, it produces fluorescence of a specific wavelength. Therefore, it can be determined whether the characteristics of the sampled fluorescence spectrum function match the characteristics of the fluorescence of the specific wavelength produced by the preset moldy product to determine whether the moldy product exists, and whether the content of the moldy product exceeds the standard.

[0054] According to an embodiment of the present invention, obtaining a mold degree quality score according to the sampled fluorescence spectrum function and the fluorescence spectrum wavelength includes: determining the mold degree quality score according to formula (3): ,

[0055] (3),

[0056] in, is the fluorescence spectrum wavelength of the t-th mildew product, is the preset fluorescence intensity of the tth mildew product, is the sampling fluorescence spectrum function of the i-th tobacco leaf sample, is the preset weight of the t-th moldy product, is the number of types of moldy products, if is the conditional function, min is the minimum value function, t≤ , i≤n, and t, , i and n are all positive integers.

[0057] According to one embodiment of the present invention, in formula (3), the conditional function Can be expressed in In the case of , otherwise it is 1. Among them, It represents the light intensity value obtained by substituting the fluorescence spectrum wavelength of the t-th moldy product into the sampling fluorescence spectrum function of the i-th tobacco leaf sample. If the light intensity value is higher than the preset fluorescence intensity of the t-th moldy product, it indicates that the content of the t-th moldy product exceeds the standard. Then the conditional function value can be set to , otherwise, if the content of the t-th moldy product does not exceed the standard, the conditional function value is 1. middle, The larger the value is, the more the content of the t-th moldy product exceeds the standard, and the lower the value of the fraction is. Therefore, this conditional function can be used as the quality function of the t-th moldy product, that is, the greater the extent to which the t-th moldy product exceeds the standard, the smaller the value of the quality function is, and if the t-th moldy product does not exceed the standard, the quality function reaches the maximum value of 1.

[0058] According to one embodiment of the present invention, the preset weights of various moldy products can be determined according to the harmfulness of various moldy products. The more harmful the moldy products are, the higher the preset weights are. The preset weights can also be set by experts. The present invention does not limit the value of the preset weights. The quality functions corresponding to various moldy products on the ith tobacco leaf sample can be weighted and summed by the preset weights to obtain the quality status of the ith tobacco leaf sample in terms of the degree of mold. The minimum value of the quality status of each tobacco leaf sample in terms of the degree of mold is taken as the quality score of the degree of mold, which can represent the worst quality status of the batch of tobacco leaf samples in terms of the degree of mold, and can be used to represent the overall quality status of the batch of tobacco leaf samples in terms of the degree of mold, that is, the higher the quality score of the degree of mold, the better the overall quality status of the batch of tobacco leaf samples in terms of mold, and the better the control of the content of various preset moldy products.

[0059] In this way, the light intensity of the fluorescence spectrum function at the wavelength can be determined by substituting the fluorescence spectrum wavelength of the moldy product into the sampled fluorescence spectrum function, thereby determining whether the moldy product exceeds the standard, and then setting the quality function of the moldy product through the conditional function. Furthermore, the quality status of the tobacco leaf sample in terms of moldiness can be solved by weighted summation, and then the minimum value of the corresponding quality status of each tobacco leaf sample in terms of moldiness can be taken to describe the overall quality status of the batch of tobacco leaf samples in terms of the control of the content of moldy products, thereby improving the accuracy and objectivity of the quality score of the degree of moldiness.

[0060] According to one embodiment of the present invention, after obtaining the conventional chemical component quality score, solid non-tobacco impurity quality score and mildew degree quality score of the batch of tobacco leaf samples, the three scores can be weighted and summed to obtain the processing quality score of the tobacco leaf sample group to be tested, thereby describing the quality status in three aspects through the detection of three spectral data, and improving the comprehensiveness and objectivity of the quality description. Furthermore, based on the processing quality score of the tobacco leaf sample group to be tested, the comprehensive quality information of tobacco leaf processing can be obtained, which can be used to describe the overall processing quality of tobacco leaves and the quality status in multiple dimensions, so that the staff can fully understand the processing quality of tobacco leaves.

[0061] According to the tobacco processing quality detection method based on hyperspectral technology of the embodiment of the present invention, a specific transmission belt can be used to carry tobacco samples, so that the transparent module and the black module are used to provide different light sources with the environment required for detecting hyperspectral data, thereby improving the detection accuracy of hyperspectral data. And different types of hyperspectral data can be detected by controlling different light sources to illuminate tobacco samples, without changing the transmission belt and the light source, the irradiation of multiple light sources and the data collection and analysis of multimodal hyperspectral can be realized, thereby determining the various quality characteristics of tobacco samples, and improving the comprehensiveness, accuracy and flexibility of quality detection. When determining the quality score of conventional chemical components, the minimum level of the quality of the batch of tobacco samples in terms of conventional chemical components can be described by the maximum value of the relative difference between the sampling transmission spectrum function and the quality standard transmission spectrum function, so as to represent the overall quality level of the batch of tobacco samples in terms of conventional chemical components, thereby representing the overall closeness of the batch of tobacco samples in terms of conventional chemical components to the standard conventional chemical components, and obtaining the quality score of conventional chemical components, which can objectively and accurately describe the quality characteristics of the batch of tobacco samples in terms of conventional chemical components. When determining the quality score of surface solid non-tobacco debris, the conditional function can be used to find the minimum and maximum points in the reflection spectrum data, and record the corresponding wavelengths, and then determine whether the reflection spectrum sampling data has the characteristics of the reflection peak and absorption peak of specific solid non-tobacco debris by cosine similarity, so as to determine whether there is specific solid non-tobacco debris on the surface of the tobacco leaf sample, and improve the accuracy of solid non-tobacco debris detection. In addition, the quality of the tobacco leaf sample with the most serious residual solid non-tobacco debris can be used to represent the overall quality of the batch of tobacco leaves in terms of surface solid non-tobacco debris, so as to improve the accuracy and objectivity of the quality score of solid non-tobacco debris. When determining the quality score of the degree of mold, the fluorescence spectrum wavelength of each type of moldy product can be substituted into the sampled fluorescence spectrum function to determine the light intensity of the fluorescence spectrum function at this wavelength, so as to determine whether the moldy product exceeds the standard, and then set the quality function of this type of moldy product through the conditional function. Furthermore, the quality status of tobacco leaf samples in terms of mildew can be solved by weighted summation, and then the minimum value of the corresponding quality status of each tobacco leaf sample in terms of mildew can be taken to describe the overall quality status of the batch of tobacco leaf samples in terms of the control of the content of mildew products, thereby improving the accuracy and objectivity of the mildew degree quality score.

[0062] Figure 4A block diagram of a tobacco processing quality detection system based on hyperspectral technology according to an embodiment of the present invention is exemplarily shown, wherein the system comprises: a first fixing module, used to fix each tobacco leaf sample in a tobacco leaf sample group to be detected on a transparent module of a conveyor belt, wherein the conveyor belt comprises a plurality of alternately distributed transparent modules and black modules, and the sizes and quantities of the transparent modules and the black modules are the same, the transparent modules are carriers made of transparent flexible materials, the black modules are carriers made of black opaque flexible materials, and a plurality of alternately distributed transparent modules and black modules are interconnected to form the conveyor belt; a first conveyor belt The driving module is used to control the movement of the conveyor belt and move each transparent module to the detection position, wherein the detection position includes a transmission light box, an LED fill light, a halogen tungsten lamp, an ultraviolet light source and a hyperspectral imager, the transmission light box is located directly below the conveyor belt and is used to emit a narrow band transmission light beam directly above, the LED fill light is located obliquely above the conveyor belt and is used to fill light for the tobacco leaf sample, the halogen tungsten lamp is located on an annular lamp holder, the annular lamp holder is located directly above the conveyor belt, the halogen tungsten lamp is used to irradiate the tobacco leaf sample with light of a preset wavelength band, and the ultraviolet light source is located obliquely above the conveyor belt and is used to emit ultraviolet light to irradiate the tobacco leaf sample In the invention, the hyperspectral imager is located directly above the annular light stand, and the transmitted light beam is received by the hyperspectral imager after passing through the transparent module, the tobacco leaf sample and the annular light stand; the first acquisition module is used to keep the LED fill light, the halogen tungsten lamp and the ultraviolet light source turned off, keep the transmission light box turned on, and when each transparent module moves to the detection position, obtain the transmission spectrum data of the tobacco leaf sample through the hyperspectral imager; the second fixing module is used to fix each tobacco leaf sample on the black module of the conveyor belt after obtaining the transmission spectrum data of each tobacco leaf sample in the tobacco leaf sample group to be detected; the second motion module is used to control the transmission The conveyor belt moves and moves each black module to the detection position; the light source control module is used to keep the transmission light box closed and keep the LED fill light turned on, and when each black module moves to the detection position, the halogen tungsten lamp and the ultraviolet light source are turned on in sequence; the second acquisition module is used to obtain the reflection spectrum data of the tobacco leaf sample through the hyperspectral imager when the halogen tungsten lamp is turned on, and to obtain the fluorescence spectrum data of the tobacco leaf sample through the hyperspectral imager when the ultraviolet light source is turned on; the processing comprehensive quality information module is used to determine the comprehensive quality information of tobacco leaf processing according to the transmission spectrum data, reflection spectrum data and fluorescence spectrum data of multiple tobacco leaf samples.

[0063] It should be understood by those skilled in the art that the embodiments of the present invention described above and shown in the accompanying drawings are only examples and do not limit the present invention. The purpose of the present invention has been fully and effectively achieved. The functional and structural principles of the present invention have been demonstrated and explained in the embodiments, and the embodiments of the present invention may be deformed or modified in any way without departing from the principles.

Claims

1. A tobacco leaf processing quality detection method based on hyperspectral technology, characterized in that: include: Fix each tobacco leaf sample in the tobacco leaf sample group to be detected on a transparent module of a conveyor belt, wherein the conveyor belt includes a plurality of alternately distributed transparent modules and black modules, and the sizes and quantities of the transparent modules and the black modules are the same, the transparent module is a carrier made of a transparent flexible material, the black module is a carrier made of a black opaque flexible material, and a plurality of alternately distributed transparent modules and black modules are interconnected to form the conveyor belt; control the movement of the conveyor belt, and move each transparent module to a detection position, wherein the detection position includes a transmission light box, an LED fill light, a halogen tungsten lamp, an ultraviolet light source and a hyperspectral imager, the transmission light box is located directly below the conveyor belt, and is used to emit a narrow band transmission light beam directly above, the LED fill light is located obliquely above the conveyor belt, and is used to fill light for the tobacco leaf samples, the halogen tungsten lamp is located on a ring light stand, and the ring light stand is located directly above the conveyor belt, the halogen tungsten lamp is used to irradiate the tobacco leaf samples with light of a preset band, and the ultraviolet light source is located obliquely above the conveyor belt, and is used to emit ultraviolet light to irradiate the tobacco leaf samples, The hyperspectral imager is located directly above the annular light stand, and the transmitted light beam is received by the hyperspectral imager after passing through the transparent module, the tobacco leaf sample and the annular light stand; the LED fill light, the halogen tungsten lamp and the ultraviolet light source are kept turned off, the transmission light box is kept turned on, and when each transparent module moves to the detection position, the transmission spectrum data of the tobacco leaf sample is obtained through the hyperspectral imager; after obtaining the transmission spectrum data of each tobacco leaf sample in the tobacco leaf sample group to be detected, each tobacco leaf sample is fixed on the black module of the conveyor belt respectively; the conveyor belt movement is controlled, and each black module is moved to the detection position; the transmission light box is kept closed and the LED fill light is kept turned on, and when each black module moves to the detection position, the halogen tungsten lamp and the ultraviolet light source are turned on in sequence; when the halogen tungsten lamp is turned on, the reflection spectrum data of the tobacco leaf sample is obtained through the hyperspectral imager, and when the ultraviolet light source is turned on, the fluorescence spectrum data of the tobacco leaf sample is obtained through the hyperspectral imager; according to the transmission spectrum data, reflection spectrum data and fluorescence spectrum data of multiple tobacco leaf samples, the comprehensive quality information of tobacco leaf processing is determined.

2. The method for detecting tobacco leaf processing quality based on hyperspectral technology according to claim 1, characterized in that: The comprehensive processing quality information of the tobacco leaf sample group to be tested is determined according to the transmission spectrum data, reflection spectrum data and fluorescence spectrum data of multiple tobacco leaf samples, including: obtaining the quality score of the conventional chemical components of the tobacco leaf sample group to be tested according to the transmission spectrum data of multiple tobacco leaf samples; obtaining the quality score of the solid non-tobacco impurities of the tobacco leaf sample group to be tested according to the reflection spectrum data of multiple tobacco leaf samples; obtaining the quality score of the degree of mildew of the tobacco leaf sample group to be tested according to the fluorescence spectrum data of multiple tobacco leaf samples; obtaining the quality score of the tobacco leaf sample group to be tested according to the quality score of the conventional chemical components, the quality score of the solid non-tobacco impurities and the quality score of the degree of mildew; and determining the comprehensive processing quality information of the tobacco leaves according to the processing quality score of the tobacco leaf sample group to be tested.

3. The method for detecting tobacco leaf processing quality based on hyperspectral technology according to claim 2, characterized in that: According to the transmission spectrum data of multiple tobacco leaf samples, the conventional chemical component quality score of the tobacco leaf sample group to be tested is obtained, including: sampling the transmission spectrum data of each tobacco leaf sample to obtain transmission spectrum sampling data; fitting the transmission spectrum sampling data to obtain a sampled transmission spectrum function; according to the preset quality grade of the tobacco leaf sample group to be tested, determining the standard transmission spectrum data corresponding to the preset quality grade; fitting the standard transmission spectrum data to obtain a quality standard transmission spectrum function; according to the sampled transmission spectrum function and the quality standard transmission spectrum function, the conventional chemical component quality score is obtained.

4. The method for detecting tobacco leaf processing quality based on hyperspectral technology according to claim 3 is characterized in that: According to the sample transmission spectrum function and the quality standard transmission spectrum function, a conventional chemical component quality score is obtained, including: According to the formula , Obtain quality scores for common chemical components ,in, is the sampling transmission spectrum function of the i-th tobacco leaf sample, is the mass standard transmission spectrum function, is the starting wavelength of the mass standard transmission spectrum function, is the end wavelength of the quality standard transmission spectrum function, n is the number of tobacco leaf samples, max is the maximum value function, is the wavelength variable, i≤n, and i and n are both positive integers.

5. The method for detecting tobacco leaf processing quality based on hyperspectral technology according to claim 2, characterized in that: According to the reflectance spectrum data of multiple tobacco leaf samples, a solid non-tobacco impurity quality score of the tobacco leaf sample group to be tested is obtained, including: sampling the reflectance spectrum data of each tobacco leaf sample to obtain reflectance spectrum sampling data; setting a first wavelength corresponding to the absorption peak and a second wavelength corresponding to the reflection peak of multiple solid non-tobacco impurities; and determining the solid non-tobacco impurity quality score according to the reflectance spectrum sampling data, the first wavelength and the second wavelength.

6. The method for detecting tobacco leaf processing quality based on hyperspectral technology according to claim 5, characterized in that: Determining a solid non-smoking debris quality score according to the reflection spectrum sampling data, the first wavelength, and the second wavelength includes: according to the formula , determine the quality score of solid non-smoking debris ,in, is the relative illumination data of the reflection spectrum sampling data of the i-th tobacco leaf sample at the j-th sampling point, is the relative illumination data of the reflectance spectrum sampling data of the i-th tobacco leaf sample at the j+1-th sampling point, is the relative illumination data of the reflection spectrum sampling data of the ith tobacco leaf sample at the j-1th sampling point, m is the number of sampling points for sampling the reflection spectrum data, and is the logical operator of "and", for The transposed vector of is the first wavelength corresponding to the kth solid non-smoke debris, is the second wavelength corresponding to the kth solid non-smoking debris, is the wavelength corresponding to the jth sampling point, is the preset threshold, is the number of types of solid non-tobacco debris, n is the number of tobacco leaf samples, max is the maximum value function, if is the conditional function, i≤n, k≤ , 2≤j≤m-1, and i, n, k, , j and m are all positive integers.

7. The method for detecting tobacco leaf processing quality based on hyperspectral technology according to claim 2, characterized in that: According to the fluorescence spectrum data of multiple tobacco leaf samples, a quality score of the degree of mildew of the tobacco leaf sample group to be tested is obtained, including: sampling the fluorescence spectrum data of each tobacco leaf sample to obtain fluorescence spectrum sampling data; fitting the fluorescence spectrum sampling data to obtain a sampled fluorescence spectrum function; determining the fluorescence spectrum wavelength of a preset mildew product; and obtaining a quality score of the degree of mildew according to the sampled fluorescence spectrum function and the fluorescence spectrum wavelength.

8. The method for detecting tobacco leaf processing quality based on hyperspectral technology according to claim 7, characterized in that: According to the sampled fluorescence spectrum function and the fluorescence spectrum wavelength, a mold degree quality score is obtained, including: according to the formula , determine the quality score of the degree of mildew ,in, is the fluorescence spectrum wavelength corresponding to the t-th mildew product, is the preset fluorescence intensity of the tth mildew product, is the sampling fluorescence spectrum function of the i-th tobacco leaf sample, is the preset weight of the t-th moldy product, is the number of types of moldy products, if is the conditional function, min is the minimum value function, t≤ , i≤n, and t, , i and n are all positive integers.

9. A tobacco leaf processing quality detection system based on hyperspectral technology for executing the method according to any one of claims 1 to 8, characterized in that: include: The first fixing module is used to fix each tobacco leaf sample in the tobacco leaf sample group to be detected on the transparent module of the conveyor belt, wherein the conveyor belt includes a plurality of alternately distributed transparent modules and black modules, and the size and quantity of the transparent modules and the black modules are the same, the transparent module is a carrier made of transparent flexible material, and the black module is a carrier made of black opaque flexible material, and a plurality of alternately distributed transparent modules and black modules are interconnected to form the conveyor belt; the first moving module is used to control the movement of the conveyor belt and move each transparent module to the detection position, wherein, The detection position includes a transmission light box, an LED fill light, a halogen tungsten lamp, an ultraviolet light source and a hyperspectral imager. The transmission light box is located directly below the conveyor belt and is used to emit a narrow band transmission light beam directly above. The LED fill light is located obliquely above the conveyor belt and is used to fill light for tobacco leaf samples. The halogen tungsten lamp is located on an annular light stand, and the annular light stand is located directly above the conveyor belt. The halogen tungsten lamp is used to irradiate tobacco leaf samples with light of a preset wavelength band. The ultraviolet light source is located obliquely above the conveyor belt and is used to emit ultraviolet light to irradiate tobacco leaf samples. The hyperspectral imager is located directly above the annular light stand. After passing through the transparent module, the tobacco leaf sample and the annular light stand, the transmitted light beam is received by the hyperspectral imager; the first acquisition module is used to keep the LED fill light, the halogen tungsten lamp and the ultraviolet light source turned off, keep the transmission light box turned on, and when each transparent module moves to the detection position, the transmission spectrum data of the tobacco leaf sample is obtained through the hyperspectral imager; the second fixing module is used to fix each tobacco leaf sample on the black module of the conveyor belt after obtaining the transmission spectrum data of each tobacco leaf sample in the tobacco leaf sample group to be detected; the second movement module is used to control the movement of the conveyor belt and fix each black The module moves to the detection position; the light source control module is used to keep the transmission light box closed and keep the LED fill light turned on, and turn on the halogen tungsten lamp and the ultraviolet light source in sequence when each black module moves to the detection position; the second acquisition module is used to obtain the reflection spectrum data of the tobacco leaf sample through the hyperspectral imager when the halogen tungsten lamp is turned on, and to obtain the fluorescence spectrum data of the tobacco leaf sample through the hyperspectral imager when the ultraviolet light source is turned on; the processing comprehensive quality information module is used to determine the comprehensive quality information of tobacco leaf processing according to the transmission spectrum data, reflection spectrum data and fluorescence spectrum data of multiple tobacco leaf samples.

Images