An intelligent imaging sorting robot for Chinese medicinal materials based on short-wave infrared multi-spectral fusion

Abstract

The present invention provides a short-wave infrared-based multi-spectral fusion intelligent imaging sorting robot for Chinese medicinal materials, including a short-wave infrared detector, an infrared coupling lens, an intelligent switching filter wheel for Chinese medicinal materials, a filter shading plate, a continuous spectrum infrared light source array, a beat alignment conveyor belt, a medicinal material sorting robot arm, a virtual instrument intelligent decision center, and a MEMS (micro-electromechanical system) spectrum probe. The present invention optimizes the decision weights of the onnx library of different spectral bands, achieving a 98% accuracy rate in identifying Chinese medicinal material components. It solves the pain point that traditional machine vision methods are difficult to distinguish between Chinese medicinal materials of the same morphology and color, and achieves a high detection beat of 0.9 seconds per package of medicines.

Description

Technical Field

[0001] The present invention belongs to the field of intelligent robots, and in particular relates to a short-wave infrared-based multi-spectral fusion intelligent imaging sorting robot for Chinese medicinal materials. Background Art

[0002] As people pay more and more attention to health and natural therapies, the market for Chinese herbal medicines is showing a booming growth trend. However, the sorting and identification of Chinese herbal medicines has always been a tedious and complex challenge. The traditional manual sorting method is not only inefficient, but also easily interfered by human factors, which in turn affects the accuracy of the sorting results.

[0003] Many Chinese medicinal materials are often very similar in appearance, such as Astragalus, Fritillaria thunbergii, stir-fried yam, Radix Trichosanthis, and Rhizoma Dioscoreae. The subtle differences between them make it difficult for traditional visual inspection methods to effectively distinguish them, which can easily lead to misjudgment or confusion. Summary of the invention

[0004] Purpose of the invention: The technical problem to be solved by the present invention is to provide a short-wave infrared multi-spectral fusion Chinese medicinal materials intelligent imaging sorting robot in view of the deficiencies of the prior art, including a short-wave infrared detector, an infrared coupling lens, a Chinese medicinal materials intelligent switching filter wheel, a filter shielding plate, a continuous spectrum infrared light source array, a beat alignment conveyor belt, a medicinal materials sorting robot arm, a virtual instrument intelligent decision-making center and a MEMS (micro-electromechanical system) spectral probe;

[0005] The Chinese medicinal materials intelligent switching filter wheel includes a light inlet;

[0006] The objective lens of the infrared coupling lens is embedded in the light entrance of the intelligent switching filter wheel of traditional Chinese medicine.

[0007] The continuous spectrum infrared light source array is tilted 30° and integrated under the filter shading plate. The preliminary spectral data of the sample is collected by the MEMS spectral probe and transmitted to the virtual instrument intelligent decision center. The virtual instrument intelligent decision center analyzes the spectral characteristics of the sample and preliminarily predicts an optimal wavelength range and corresponding filter through the peak value of the spectral characteristic diagram (reflectivity greater than 60%). Subsequently, based on further identification and analysis of the components of Chinese medicinal materials, the virtual instrument intelligent decision center determines the precise wavelength required and instructs the Chinese medicinal materials intelligent switching filter wheel to accurately switch the matching filter to the light inlet.

[0008] The current infrared band signal of the Chinese medicinal materials is collected by the short-wave infrared detector above the eyepiece of the infrared coupling lens.

[0009] The short-wave infrared detector is a 640×512 array infrared detector. Each short-wave infrared detection pixel in the short-wave infrared detector can use the filter of the intelligent switching filter wheel of Chinese medicinal materials to collect the multi-spectral characteristic signals of Chinese medicinal materials; the virtual instrument intelligent decision center is used to combine all pixels, which can not only realize the scanning of Chinese medicinal materials components, but also construct the spatial spectrum imaging of the infrared characteristic shapes of Chinese medicinal materials. The target surface of the short-wave infrared detector adopts an array layout. In view of the complex components, diverse shapes and more impurities of Chinese medicinal materials, the array short-wave infrared detector can realize the scanning of Chinese medicinal materials components and construct the spatial spectrum imaging of the infrared characteristic shapes of multi-dimensional Chinese medicinal materials.

[0010] The MEMS spectral probe characterizes Chinese medicinal materials with similar appearance and color to Astragalus, Rhizoma Dioscoreae, and fried Chinese yam, as well as Chinese medicinal materials with problems such as mold and bad core. In the infrared band, the characteristics of 1173nm, 1250nm, 1327nm, 1450nm, 1519nm, and 1605nm are relatively strong, so these wavelengths are selected in the intelligent switching filter wheel for Chinese medicinal materials. If other Chinese medicinal materials are to be identified, the MEMS spectral probe can be used to select other infrared wavelength filters and install them in the intelligent switching filter wheel for Chinese medicinal materials.

[0011] After receiving the command from the virtual instrument intelligent decision-making center, the next package of Chinese medicinal materials to be sorted is transmitted to the bottom of the light entrance through the beat alignment conveyor belt. The infrared spectrum characteristic signal of the Chinese medicinal materials enters the objective lens of the infrared coupling lens through the incident port and is collected by the short-wave infrared detector to realize infrared spectrum imaging; the virtual instrument intelligent decision-making center controls the uniform movement of the intelligent switching filter wheel of the Chinese medicinal materials. According to the wavelength of the filter switched to the light entrance, the virtual instrument intelligent decision-making center switches the spectrum deep learning onnx library of the wavelength in real time, gives the composition and quality information of the Chinese medicinal materials fused by multi-spectrum, and issues the command to detect the next package of Chinese medicinal materials by beat alignment conveyor belt, and at the same time assigns sorting mark instructions to the medicinal materials sorting robot arm.

[0012] The virtual instrument intelligent decision center is used to realize infrared spectrum feature recognition and control the movement of various components of the sorting robot, including the following steps:

[0013] Step 1: Collect Chinese medicinal materials with known ingredients using the MEMS spectral probe, such as Chinese medicinal materials with similar appearance colors such as Astragalus, Rhizoma Dioscoreae, and Rhizoma Dioscoreae, and add characteristic labels to Chinese medicinal materials with problems such as mold and rotten core; select six infrared wavelength filters according to the infrared characteristic peak intensity of the Chinese medicinal materials, and fit the identification decision weight of each Chinese medicinal material label (including labels of problematic medicinal materials such as mold and rotten core) when multi-wavelength fusion is used;

[0014] Step 2, select the six infrared wavelength filters obtained in step 1 to form an intelligent switching filter wheel for Chinese medicinal materials, use a short-wave infrared detector to collect the spectral signals of the labeled Chinese medicinal materials (including Chinese medicinal materials with moldy, bad core and other problems) and image them, to form a data set of spectral images of Chinese medicinal materials, and use the data set of spectral images of Chinese medicinal materials at six infrared wavelengths to obtain the label features of Chinese medicinal materials with known ingredients through YOLOv8 model training, to form an onnx Chinese medicinal material component library for multi-spectral recognition of six infrared wavelengths;

[0015] Step 3, when sorting the unknown Chinese medicinal material packages, the beat is aligned with the conveyor belt to send the Chinese medicinal material packages to the bottom of the light inlet in turn, based on the wavelength information at the light inlet, and referring to the wavelength information, the data of each Chinese medicinal material label (including labels of problematic medicinal materials such as moldy and spoiled) stored in the onnx Chinese medicinal material component library for multispectral identification of six infrared wavelengths, each package of medicinal materials is dynamically identified, and the confidence of each Chinese medicinal material label (including labels of problematic medicinal materials such as moldy and spoiled) under each wavelength channel is dynamically identified within 100ms;

[0016] In step 4, the virtual instrument intelligent decision center combines the Chinese medicinal material label recognition decision weight obtained in step 1 with the confidence obtained in step 3 to obtain the medicinal material composition and quality in the Chinese medicinal material package with unknown ingredients; controls the beat to align with the conveyor belt, and conveys the Chinese medicinal material packages with mis-dispensing, mildew, and bad quality problems to the medicinal material sorting robot arm for picking out and manual processing; normal medicine packages are sent to the designated medicine output area, and the next package of Chinese medicinal materials to be sorted is transported to the bottom of the light inlet for inspection.

[0017] Step 4 includes:

[0018] The MEMS spectral probe collects infrared spectral data of different Chinese medicinal materials in the wavelength range of 1100nm to 1700nm. In the infrared spectral data, the horizontal axis is the wavelength and the vertical axis is the reflectivity. The infrared spectral data is mapped to a high-dimensional feature space through nonlinear mapping, and an optimal hyperplane space is found to obtain spatial spectral information of different categories of Chinese medicinal materials. In order to determine the contribution of each spectral data in the fusion result, the expression of the Chinese medicinal material label recognition decision weight vector w is:

[0019] w=[w 1 ,w 2 ,..,w m ]

[0020] Among them, the elements in the vector represent the contribution weight of each short-wave infrared wavelength to the identification of Chinese medicinal materials, m represents the number of infrared wavelength bands; w m Represents the contribution weight of the m-band shortwave infrared wavelength to the identification of Chinese medicinal materials.

[0021] In step 4, the contribution weight w of the k-band short-wave infrared wavelength to the identification of Chinese medicinal materials k The calculation formula is:

[0022]

[0023] Among them, k ranges from 1 to m, n represents the number of medicinal material samples in the Chinese medicinal material spectral dataset, and α i represents the Lagrange multiplier of the i-th Chinese medicinal material sample, x ik represents the spectral characteristic ratio of the i-th Chinese medicinal material sample under the k-band short-wave infrared, y i Represents the i-th Chinese medicinal material category label.

[0024] The spectral deep learning onnx library of each Chinese medicinal material at the corresponding wavelength of the light entrance needs to be synchronized with the Chinese medicinal material intelligent switching filter wheel of the mechanical device, that is, the filter wavelength in the Chinese medicinal material intelligent switching filter wheel rotated to the light entrance is the same as the wavelength of the spectral deep learning onnx library in the virtual instrument intelligent decision-making center.

[0025] The present invention has the following technical effects: by integrating multiple advanced technologies such as image recognition, machine learning, and spectral analysis, the present invention successfully solves two major problems in the sorting of Chinese medicinal materials, namely, the distinction between similar medicinal materials and the identification of moldy medicinal materials, and on this basis achieves a significant improvement in sorting efficiency. The present invention improves the automation and intelligence level of Chinese medicinal materials sorting. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 It is the overall structure diagram of the device of the present invention.

[0027] Figure 2 It is the spectral reflectance diagram of the Chinese medicinal material of the present invention.

[0028] Figure 3 It is a bottom view of the device of the present invention.

[0029] Figure 4 These are six single-band infrared images of the Chinese medicinal materials described in the present invention.

[0030] Figure 5 This is a physical diagram of the device of the present invention.

[0031] Figure 6 This is a result fusion diagram of the present invention.

[0032] Explanation of the accompanying drawings: 1 is a short-wave infrared detector, 2 is an infrared coupling lens, 3 is an intelligent switching filter wheel for Chinese medicinal materials, 4 is a filter shading plate, 5 is a continuous spectrum infrared light source array, 6 is a beat alignment conveyor belt, 7 is a medicinal material sorting robot arm, 8 is a virtual instrument intelligent decision center; 9 is a light inlet; 10 is a MEMS spectral probe. DETAILED DESCRIPTION

[0033] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments, and the above and / or other advantages of the present invention will become more clear.

[0034] like Figure 1 , Figure 5 As shown, the embodiment of the present invention provides a short-wave infrared-based multi-spectral fusion Chinese medicinal material intelligent imaging sorting robot, including a short-wave infrared detector 1, an infrared coupling lens 2, a Chinese medicinal material intelligent switching filter wheel 3, a filter shading plate 4, a continuous spectrum infrared light source array 5, a beat alignment conveyor belt 6, a medicinal material sorting robot arm 7, a virtual instrument intelligent decision center 8 and a MEMS spectrum probe 10;

[0035] The Chinese medicinal material intelligent switching filter wheel 3 includes a light inlet 9;

[0036] The objective lens of the infrared coupling lens 2 is embedded in the light inlet 9 of the Chinese medicinal material intelligent switching filter wheel 3 .

[0037] The continuous spectrum infrared light source array 5 is tilted 30° and integrated below the filter shading plate 4. The preliminary spectral data of the sample is collected by the MEMS spectral probe 10 and transmitted to the virtual instrument intelligent decision center 8. The virtual instrument intelligent decision center 8 analyzes the spectral characteristics of the sample and predicts the filter corresponding to the best wavelength range. The Chinese medicinal material intelligent switching filter wheel 3 identifies the required wavelength according to the current Chinese medicinal material components of the virtual instrument intelligent decision center 8, and accurately switches the wavelength filter to the light inlet 9. The short-wave infrared detector 1 collects the current infrared band signal of the Chinese medicinal material. The short-wave infrared detector 1 is placed above the eyepiece of the infrared coupling lens 2, and its detector target surface adopts an array layout. In view of the complex components, diverse shapes and more impurities of Chinese medicinal materials, the array-type short-wave infrared detector 1 can not only realize the scanning of Chinese medicinal material components, but also construct the spatial spectrum imaging of the infrared characteristic shape of Chinese medicinal materials;

[0038] After receiving the command from the virtual instrument intelligent decision center 8, the next package of Chinese medicinal materials to be sorted is transmitted to the bottom of the light inlet 9 through the beat alignment conveyor belt 6. The infrared spectrum characteristic signal of the Chinese medicinal materials enters the objective lens of the infrared coupling lens 2 through the incident port and is collected by the short-wave infrared detector 1 to realize infrared spectrum imaging; the virtual instrument intelligent decision center 8 controls the uniform movement of the Chinese medicinal materials intelligent switching filter wheel 3. The virtual instrument intelligent decision center 8 switches the spectrum deep learning onnx library of the wavelength in real time according to the wavelength of the filter switched to the light inlet 9, and gives the Chinese medicinal materials composition and quality information of multi-spectral fusion. And issue a command to detect the next package of Chinese medicinal materials by the beat alignment conveyor belt 6, and at the same time assign sorting mark instructions to the medicinal materials sorting mechanical arm 7.

[0039] The virtual instrument intelligent decision center 8 is used to realize infrared spectrum feature recognition and control the movement of various components of the sorting robot, including the following steps:

[0040] Step 1, using the MEMS spectral probe 10 to collect Chinese medicinal materials with known ingredients, such as Chinese medicinal materials with similar appearance colors such as astragalus, rhizoma dioscoreae, and fried yam, as well as Chinese medicinal materials with problems such as mold and rotten core, and label them with characteristic labels; select six infrared wavelength filters according to the infrared characteristic peak intensity of the Chinese medicinal materials, and fit the identification decision weight of each Chinese medicinal material label when multi-wavelength fusion is performed;

[0041] Step 2, select the six infrared characteristic wavelength filters obtained in step 1 to form a Chinese medicinal material intelligent switching filter wheel 3, use the short-wave infrared detector 1 to collect the spectral signal of the labeled Chinese medicinal materials and image it, so as to form a data set of spectral images of Chinese medicinal materials, and use the data set of spectral images of Chinese medicinal materials under six infrared characteristic wavelengths to obtain the label features of each known component of Chinese medicinal materials through YOLOv8 model training, so as to form an onnx Chinese medicinal material component library for multi-spectral recognition with six wavelengths;

[0042] Step 3, when sorting the unknown Chinese medicinal material packages, the beat is aligned with the conveyor belt 6 to send the Chinese medicinal material packages to the bottom of the light inlet 9 in sequence. Each package of medicinal material is dynamically identified within 100ms according to the wavelength at the light inlet 9 and the onnx component library of the spectral imaging of each labeled Chinese medicinal material under the corresponding wavelength.

[0043] In step 4, the virtual instrument intelligent decision center 8 combines the Chinese medicinal material label recognition decision weight obtained in step 1 with the confidence obtained in step 3 to obtain the medicinal material composition and quality in the Chinese medicinal material package with unknown ingredients; controls the beat to align the conveyor belt 6 to convey the Chinese medicinal material packages with mis-dispensing, mildew and bad quality problems to the medicinal material sorting robot 7 to be picked out and handed over to manual processing; normal medicine packages are sent to the designated medicine output area, and the next package of Chinese medicinal materials to be sorted is transported to the bottom of the light inlet 9 for inspection.

[0044] The short-wave infrared detector 1 is a 640×512 array infrared detector. Each short-wave infrared detection pixel in the short-wave infrared detector 1 can utilize the filter of the intelligent switching filter wheel 3 of the Chinese medicinal materials to collect the multi-spectral characteristic signals of the Chinese medicinal materials; the virtual instrument intelligent decision center 8 is used to combine all pixels, which can not only realize the scanning of the components of the Chinese medicinal materials, but also construct the spatial spectrum imaging of the infrared characteristic shapes of the Chinese medicinal materials.

[0045] The MEMS spectroscopic probe 10 characterizes Chinese medicinal materials with similar appearance and color to Astragalus, Rhizoma Dioscoreae, and Chinese yam, as well as Chinese medicinal materials with moldy and bad core problems. In the infrared band, the characteristics of 1173nm, 1250nm, 1327nm, 1450nm, 1519nm, and 1605nm are relatively strong, so these wavelengths are selected in the Chinese medicinal material intelligent switching filter wheel 3. If other Chinese medicinal materials are to be identified, the MEMS spectroscopic probe 10 can be used to select other infrared wavelength filters and install them in the Chinese medicinal material intelligent switching filter wheel 3.

[0046] The spectral deep learning onnx library of each Chinese medicinal material at the corresponding wavelength of the light inlet 9 needs to be synchronized with the Chinese medicinal material intelligent switching filter wheel 3 of the mechanical device, that is, the filter wavelength in the Chinese medicinal material intelligent switching filter wheel 3 rotated to the light inlet is the same as the wavelength of the spectral deep learning onnx library in the virtual instrument intelligent decision center 8.

[0047] The MEMS spectral probe 10 collects infrared spectral data of different Chinese medicinal materials in the wavelength range of 1100nm to 1700nm, where the horizontal axis is the wavelength and the vertical axis is the reflectivity; the spectral data is mapped to a high-dimensional feature space through nonlinear mapping, and an optimal hyperplane is found to achieve the classification of different types of Chinese medicinal materials. In order to determine the contribution of each spectral data in the fusion result, the expression of the Chinese medicinal material label recognition decision weight vector w is:

[0048] w=[w 1 ,w 2 ,..,w m ]

[0049] Among them, the elements in the vector represent the contribution weight of each short-wave infrared wavelength to the identification of traditional Chinese medicine, and m represents the number of infrared wavelength bands.

[0050] The kth element w in the decision weight vector of Chinese herbal medicine label recognition k The calculation formula is:

[0051]

[0052] Among them, k ranges from 1 to m, n represents the number of Chinese medicinal materials samples in the data, and α i represents the Lagrange multiplier of the i-th Chinese medicinal material sample, x ik It represents the spectral characteristic ratio of the i-th Chinese medicinal material sample under the k-band short-wave infrared, y i It represents the i-th Chinese medicinal material category label.

[0053] like Figure 2As shown, after the spectral signals of the Chinese medicinal materials are collected and quantified by the spectrometer, different wavelengths (wave numbers) are used as the horizontal coordinates and the reflectivity of the Chinese medicinal materials is used as the vertical coordinates to draw a two-dimensional intensity diagram of the reflections of different Chinese medicinal materials.

[0054] like Figure 3 As shown, the small circle is a bottom view of the continuous spectrum infrared light source array 5, and the large circle is a bottom view of the light entrance 9.

[0055] like Figure 4 As shown in the figure, the six single-band infrared imaging grayscale feature maps of Chinese medicinal materials are obtained by short-wave infrared imaging technology. These images can reflect the absorption characteristics of Chinese medicinal materials in different infrared bands and can be used for the identification of Chinese medicinal materials.

[0056] like Figure 6 As shown, after completing the identification of Chinese medicinal materials at several wavelengths, the identification shows the medicinal material components and confidence. The identification results are finally multiplied one by one with the elements in the fusion weight vector to obtain the composition of the Chinese medicinal materials.

[0057] The present invention provides a multi-spectral fusion Chinese medicinal materials intelligent imaging sorting robot based on short-wave infrared. There are many methods and ways to implement the technical solution. The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, without departing from the principle of the present invention, several improvements and modifications can be made, and these improvements and modifications should also be regarded as the protection scope of the present invention. All components not specified in this embodiment can be implemented by existing technologies.

Claims

1. A short-wave infrared multi-spectral fusion Chinese medicinal materials intelligent imaging sorting robot, characterized in that: It includes a short-wave infrared detector (1), an infrared coupling lens (2), a Chinese medicinal material intelligent switching filter wheel (3), a filter shading plate (4), a continuous spectrum infrared light source array (5), a beat alignment conveyor belt (6), a medicinal material sorting robot arm (7), a virtual instrument intelligent decision-making center (8) and a MEMS spectral probe (10); The virtual instrument intelligent decision center (8) is used to realize infrared spectrum feature recognition and control the movement of various components of the sorting robot, including the following steps: Step 1, using the MEMS spectral probe (10) to collect Chinese medicinal materials with known ingredients, and marking the problematic Chinese medicinal materials with characteristic labels; selecting six infrared wavelength filters according to the infrared characteristic peak intensity of the Chinese medicinal materials, and fitting the identification decision weight of each Chinese medicinal material label when multi-wavelength fusion is performed; Step 2, select the six infrared wavelength filters obtained in step 1 to form a Chinese medicinal material intelligent switching filter wheel (3), use the short-wave infrared detector (1) to collect the spectral signal of the labeled Chinese medicinal material and image it, so as to form a data set of spectral images of the Chinese medicinal material, and use the data set of spectral images of the Chinese medicinal material at six infrared wavelengths to obtain the label features of the Chinese medicinal material with known components through YOLOv8 model training, so as to form an ONNX Chinese medicinal material component library for multi-spectral recognition at six infrared wavelengths; Step 3, when sorting the Chinese medicinal material packages of unknown ingredients, the Chinese medicinal material packages are sequentially sent to the bottom of the light inlet (9) by aligning the beat with the conveyor belt (6), and based on the wavelength information at the light inlet (9), and referring to the Chinese medicinal material label data stored in the onnx Chinese medicinal material component library for multi-spectral identification of six infrared wavelengths under the wavelength information, each package of medicinal materials is dynamically identified, and the confidence of each Chinese medicinal material label under each wavelength channel is dynamically identified; In step 4, the virtual instrument intelligent decision center (8) combines the Chinese medicinal material label recognition decision weight obtained in step 1 with the confidence obtained in step 3 to obtain the medicinal material composition and quality in the Chinese medicinal material package with unknown ingredients; controls the beat to align with the conveyor belt (6), and conveys the Chinese medicinal material packages with mis-dispensing, mildew, and bad quality problems to the medicinal material sorting robot arm (7) to be picked out and handed over to manual processing; normal medicine packages are sent to the designated medicine output area, and the next package of Chinese medicinal materials to be sorted is transported to the bottom of the light inlet (9) for inspection.

2. According to claim 1, a short-wave infrared-based multi-spectral fusion Chinese medicinal materials intelligent imaging sorting robot is characterized in that: The Chinese medicinal material intelligent switching filter wheel (3) comprises a light inlet (9); The objective lens of the infrared coupling lens (2) is embedded in the light inlet (9) of the Chinese medicinal material intelligent switching filter wheel (3).

3. The multi-spectral fusion Chinese medicinal materials intelligent imaging sorting robot based on short-wave infrared according to claim 2 is characterized in that: The continuous spectrum infrared light source array (5) is tilted 30 degrees and integrated below the filter shading plate (4). The preliminary spectral data of the sample is collected by the MEMS spectral probe (10) and transmitted to the virtual instrument intelligent decision center (8). The virtual instrument intelligent decision center (8) analyzes the spectral characteristics of the sample and preliminarily predicts an optimal wavelength range and corresponding filter through the peak value of the spectral characteristic diagram. Subsequently, based on further identification and analysis of the components of the Chinese medicinal materials, the virtual instrument intelligent decision center (8) determines the required precise wavelength and instructs the Chinese medicinal materials intelligent switching filter wheel (3) to accurately switch the matching filter to the light inlet (9).

4. The multi-spectral fusion Chinese medicinal materials intelligent imaging sorting robot based on short-wave infrared according to claim 3 is characterized in that: The current infrared band signal of the Chinese medicinal material is collected by a short-wave infrared detector (1) above the eyepiece of the infrared coupling lens (2).

5. The multi-spectral fusion Chinese medicinal materials intelligent imaging sorting robot based on short-wave infrared according to claim 4 is characterized in that: The short-wave infrared detector (1) is a 640×512 array infrared detector. Each short-wave infrared detection pixel in the short-wave infrared detector (1) can utilize the filter of the intelligent switching filter wheel (3) of the Chinese medicinal material to collect the multi-spectral characteristic signal of the Chinese medicinal material. The virtual instrument intelligent decision center (8) is used to combine all pixels to realize the scanning of the components of the Chinese medicinal material and construct the spatial spectrum imaging of the infrared characteristic shape of the Chinese medicinal material.

6. The multi-spectral fusion Chinese medicinal materials intelligent imaging sorting robot based on short-wave infrared according to claim 5 is characterized in that: After receiving the command from the virtual instrument intelligent decision-making center (8), the next package of Chinese medicinal materials to be sorted is transmitted to the bottom of the light inlet (9) through the beat alignment conveyor belt (6), and the infrared spectrum characteristic signal of the Chinese medicinal materials enters the objective lens of the infrared coupling lens (2) through the incident port and is collected by the short-wave infrared detector (1) to realize infrared spectrum imaging; the virtual instrument intelligent decision-making center (8) controls the uniform movement of the Chinese medicinal materials intelligent switching filter wheel (3), and the virtual instrument intelligent decision-making center (8) switches the spectrum deep learning onnx library of the wavelength in real time according to the wavelength of the filter switched to the light inlet (9), provides the Chinese medicinal materials composition and quality information of multi-spectral fusion, and issues the command for the beat alignment conveyor belt (6) to detect the next package of Chinese medicinal materials, and at the same time distributes the sorting mark instruction to the medicinal materials sorting robot arm (7).

7. The multi-spectral fusion Chinese medicinal materials intelligent imaging sorting robot based on short-wave infrared according to claim 6 is characterized in that: Step 4 includes: The MEMS spectral probe (10) collects infrared spectral data of different Chinese medicinal materials within a wavelength range of 1100 nm to 1700 nm, wherein the abscissa of the infrared spectral data is the wavelength and the ordinate is the reflectivity; the infrared spectral data is mapped to a high-dimensional feature space through nonlinear mapping, and an optimal hyperplane space is searched to obtain spatial spectral information of Chinese medicinal materials of different categories, and the expression of the Chinese medicinal material label recognition decision weight vector w is: in=[in1,in2,..,in m ] Among them, the elements in the vector represent the contribution weight of each short-wave infrared wavelength to the identification of Chinese medicinal materials, m represents the number of infrared wavelength bands; w m Represents the contribution weight of the m-band shortwave infrared wavelength to the identification of Chinese medicinal materials.

8. The multi-spectral fusion Chinese medicinal materials intelligent imaging sorting robot based on short-wave infrared according to claim 7 is characterized in that: In step 4, the contribution weight w of the k-band short-wave infrared wavelength to the identification of Chinese medicinal materials k The calculation formula is: Among them, k ranges from 1 to m, n represents the number of medicinal material samples in the Chinese medicinal material spectral dataset, and α i represents the Lagrange multiplier of the i-th Chinese medicinal material sample, x ik represents the spectral characteristic ratio of the i-th Chinese medicinal material sample under the k-band short-wave infrared, y i Represents the i-th Chinese medicinal material category label.

9. The multi-spectral fusion Chinese medicinal materials intelligent imaging sorting robot based on short-wave infrared according to claim 8 is characterized in that: The spectral deep learning onnx library of each Chinese medicinal material at the corresponding wavelength of the light inlet (9) needs to be synchronized with the Chinese medicinal material intelligent switching filter wheel (3) of the mechanical device, that is, the filter wavelength in the Chinese medicinal material intelligent switching filter wheel (3) rotated to the light inlet (9) is the same as the wavelength of the spectral deep learning onnx library in the virtual instrument intelligent decision center (8).

Images