Raman peak intensity information-based genuine medicinal material structured feature selection method and system

Abstract

The invention provides a raman peak intensity information-based genuine medicinal material structured feature selection method and system. The method comprises the following steps: scanning medicinal material slices to generate two-dimensional distribution data of chemical components; a three-dimensional feature is constructed by using dual-wavelength laser irradiation and combining Raman peak intensity difference. And identifying a plurality of concentration regions and boundary cross points thereof through spatial topology analysis, establishing a spatial correlation model, analyzing a distribution rule, and finally forming component arrangement graph structure data. Key information is screened based on genuine medicinal material features, and after a discrimination model is input, a result is output in combination with spatial and spectral features, so that accurate discrimination is realized. And selecting the structured characteristics of the genuine medicinal materials based on Raman peak intensity information. According to the method, the spatial topology and the chemical component distribution characteristics of the Raman spectrum are fused to construct the graph structured identification model, so that the accuracy and the interpretability of genuine medicinal material identification are remarkably improved.

Description

Technical Field

[0001] This application relates to the technical field of structured feature selection, and in particular to a method and system for selecting structured features of genuine medicinal materials based on Raman peak intensity information. Background Art

[0002] In the field of quality control of Chinese medicinal materials, due to the ecological environment differences in specific production areas, genuine medicinal materials have formed unique spatial distribution patterns of active ingredients (such as gradient distribution, regional enrichment, etc.). Although traditional physical and chemical detection can measure the component content, it cannot characterize the spatial heterogeneity of components, resulting in insufficient discrimination accuracy. There is an urgent need for a technology that can non-destructively and quickly obtain the three-dimensional spatial distribution characteristics of multiple components inside / on the surface of medicinal materials, and establish a quantifiable spatial pattern discrimination model to support the accurate identification and traceability of genuine medicinal materials.

[0003] Currently, the mainstream solution adopts near-infrared hyperspectral imaging (Near-Infrared Hyperspectral Imaging, NIR-HSI) combined with chemometric methods. This technology collects hyperspectral cube data of medicinal material samples in the spectral range of a specific near-infrared short-wave region, and uses partial least squares discriminant analysis (Partial Least Squares Discriminant Analysis, PLS-DA) or support vector machine (Support Vector Machine, SVM) to extract characteristic bands and construct distribution heat maps of chemical components in different regions. By comparing the spatial distribution differences between samples from the genuine producing area and non-genuine producing areas (such as the annular enrichment characteristics of flavonoid components in the cross-section of rhizomes), identification based on spatial pattern matching is realized.

[0004] However, this technology has three limitations: First, the near-infrared spectrum has weak resolution ability for components with similar structures (such as isomers), resulting in limited accuracy of extracting spatial distribution characteristics; second, light scattering interference is easily generated on the surface of high-humidity medicinal materials, affecting the accuracy of three-dimensional space reconstruction; third, the existing model relies on a large number of labeled samples to construct a "standard spatial atlas library", but the acquisition cost of genuine medicinal material samples is high and there are geographical protection barriers, resulting in insufficient generalization ability of the model. These defects restrict the practical application value of this technology in the identification of rare medicinal materials. Summary of the Invention

[0005] This application provides a method and system for selecting structured features of genuine medicinal materials based on Raman peak intensity information to solve the problem of poor structured feature selection effect in the prior art.

[0006] In the first aspect, this application provides a method for selecting structured features of genuine medicinal materials based on Raman peak intensity information, including: Spatially sample the surface area of the medicinal material slices through a scanning device to generate two-dimensional distribution data containing concentration gradients of different chemical components; Alternately project the excitation light generated by a dual-wavelength laser source onto the same sampling area, generate a spatial response contrast map based on the Raman peak intensity differences obtained at two wavelengths, and form a three-dimensional feature stack according to the spatial response contrast map and the two-dimensional distribution data; Perform spatial topological analysis on the three-dimensional feature stack to identify at least three target concentration regions and their boundary intersection points; Construct a spatial association model between the target concentration regions, and analyze the distribution density and direction vector relationship of the boundary intersection points in each target concentration region through the spatial association model to generate graph structure data representing the spatial arrangement pattern of the medicinal material components; Perform a structured feature selection operation, retain the feature combinations that match the prior knowledge of the spatial distribution of authentic medicinal materials in the graph structure data based on the constraint conditions, and use the screened feature combinations as structured features to input into the authenticity discrimination model, so as to output the discrimination result according to the combined weights of the feature combinations in the spatial dimension and the spectral dimension through the discrimination model.

[0007] Optionally, the retaining the feature combinations that match the prior knowledge of the spatial distribution of authentic medicinal materials in the graph structure data based on the constraint conditions includes: Extract the feature distribution law of the graph structure data from the historical authentic medicinal material sample library, and construct a constraint rule set through the feature distribution law; Traverse all the feature combinations in the graph structure data, and screen out the feature combinations that meet the preset conditions as the primary screening set; wherein, the feature combinations that meet the preset conditions refer to the feature combinations with multiple constraint marks; For each feature combination in the primary screening set, calculate the product value of the regional area dispersion and the intersection point density, and retain the feature combinations whose product value ranks within a predefined ratio range as the screened feature combinations.

[0008] Optionally, using the screened feature combinations as structured features to input into the authenticity discrimination model, so as to output the discrimination result according to the combined weights of the feature combinations in the spatial dimension and the spectral dimension through the discrimination model, includes: Use the screened feature combinations as discrete units with spatial coordinate marks to input into the authenticity discrimination model; A dual-channel calculation path is established inside the authenticity discrimination model, in which the first channel maps the two-dimensional coordinate value of each discrete unit to a virtual plane divided by an orthogonal grid, calculates the standard deviation of the spectral response amplitude difference of all discrete units in each orthogonal grid, and forms a spatial fluctuation coefficient in grid units; the second channel extracts the straight-line distance value between each discrete unit and the boundary intersection of the target concentration area to which it belongs, and generates an angle correction factor in combination with the directional correlation strength; Generate a joint weight value according to the spatial fluctuation coefficient and the angle correction factor by an asymmetric superposition method; Construct a computing network including a ring connection structure, inject the joint weight value of each discrete unit into its corresponding ring node, realize the spatial transmission of the weight value through the impedance matching relationship between adjacent ring nodes, enable the output end of the computing network to receive the joint weight value output by each port, and compare the sum of the joint weight values of each port with the preset azimuth reference value to generate a phase comparison result; An azimuth offset map is generated according to the phase comparison result, and the spatial distribution forms of the offset extreme value areas and the target concentration areas that appear continuously in the azimuth offset map are overlapped and analyzed. When the boundary of the offset extreme value area forms a specific angle relationship with the geometric center of at least three target concentration areas, the confirmation mechanism of the authenticity discrimination model is triggered, and the identification result is output.

[0009] Optionally, the analyzing the distribution density and directional vector relationship of the boundary intersection points in each target concentration area by the spatial correlation model to generate graph structure data representing the spatial arrangement pattern of medicinal material components includes: An initial topological network is established through the spatial association model, with the geometric center of gravity of each target concentration area as a node and the boundary intersection as a connection hub; In the initial topological network, the corresponding number of intersections within a preset radius around each boundary intersection is calculated as a density distribution parameter, and the direction angle of the line connecting each boundary intersection and the center of gravity of the target concentration area to which it belongs is recorded as a direction relationship parameter; Vector synthesis of boundary intersections with overlapping density distribution parameters is performed to generate weighted edges representing the strength of interactions between regions; The density distribution parameter and the direction relationship parameter are used as node attributes, and the vector synthesis direction of the weighted edge is used as an edge attribute to construct graph structure data containing multi-level spatial association features.

[0010] Optionally, the surface area of the medicinal material slice is spatially sampled by a scanning device to generate two-dimensional distribution data containing concentration gradients of different chemical components, including: The surface area is irradiated point by point along a preset grid path by a scanning device, and each grid node corresponds to a sampling position; At each sampling position, monochromatic excitation light is vertically incident on the surface of the medicinal material, and the Raman scattering signal reflected at the sampling position is synchronously received. The intensity value in a preset wavelength range in the Raman scattering signal is extracted as the characteristic quantity of the grid node; According to the spatial coordinate arrangement of the grid nodes, the characteristic quantities of all grid nodes are filled into a two-dimensional matrix in row-column order, and the difference between the characteristic quantities of adjacent grid nodes is calculated. If the difference exceeds the set threshold, interpolation nodes are inserted into the corresponding row-column gap to supplement the characteristic quantities of the interpolation nodes to form a continuously distributed two-dimensional data plane; The characteristic quantity of each sampling position on the two-dimensional data plane is mapped to a gray value to generate two-dimensional distribution data with spatial coordinates as the horizontal and vertical axes and gray values representing the chemical component concentration.

[0011] Optionally, the excitation light generated by the dual-wavelength laser source is alternately projected onto the same sampling area, a spatial response contrast map is generated based on the Raman peak intensity difference obtained under two wavelengths, and a three-dimensional feature stack is formed according to the spatial response contrast map and the two-dimensional distribution data, including: The output wavelength of the dual-wavelength laser source is alternately switched by a synchronous controller, and a complete spectral acquisition is performed on the same sampling area at each wavelength; The intensity value of the characteristic peak of the Raman signal collected at the first wavelength is marked as the first response value, and the intensity value of the characteristic peak of the Raman signal collected at the second wavelength is marked as the second response value; The difference between the first response value and the second response value is calculated for each spatial sampling point, and the difference is mapped into the same spatial coordinate system as the two-dimensional distribution data to generate a spatial response contrast map; The spatial response contrast map is used as an additional signal layer and stacked with the two-dimensional distribution data to form a three-dimensional feature stack with three signal intensities.

[0012] Optionally, performing spatial topology analysis on the three-dimensional feature stack to identify at least three target concentration regions and their boundary intersection points, including: Performing a segmentation operation on each signal layer in the three-dimensional feature stack, and extracting a continuous region with a signal intensity value higher than a preset threshold as a candidate concentration region; Performing a spatial position comparison on the candidate concentration regions in multiple signal layers, and retaining the candidate regions with a spatial overlap degree exceeding a set ratio in at least two signal layers as target concentration regions; Detecting the boundary curves of each target concentration region, and calculating the minimum distance and curvature change rate between the boundary curves corresponding to adjacent target concentration regions; When the minimum distance between the boundary curves corresponding to adjacent target concentration regions is less than a preset ratio of the regional average width and the curvature change rate exceeds a preset mutation threshold, it is determined that there is an intersection point between the boundaries corresponding to the adjacent target concentration regions.

[0013] In a second aspect, the present application provides a structured feature selection system for authentic medicinal materials based on Raman peak intensity information, including: A first generation module for spatially sampling the surface area of a medicinal material slice through a scanning device to generate two-dimensional distribution data containing concentration gradients of different chemical components; A second generation module for alternately projecting the excitation light generated by a dual-wavelength laser source onto the same sampling area, generating a spatial response contrast map based on the Raman peak intensity differences obtained under two wavelengths, and forming a three-dimensional feature stack according to the spatial response contrast map and the two-dimensional distribution data; An identification module for performing spatial topological analysis on the three-dimensional feature stack to identify at least three target concentration regions and their boundary intersection points; A third generation module for constructing a spatial association model between the target concentration regions and analyzing the distribution density and direction vector relationship of the boundary intersection points in each target concentration region through the spatial association model to generate graph structure data representing the spatial arrangement pattern of the medicinal material components; An output module for performing structured feature selection operations, retaining, through constraint conditions, the feature combinations that match the prior knowledge of the spatial distribution of authentic medicinal materials in the graph structure data, and inputting the selected feature combinations as structured features into an authenticity discrimination model, so as to output a discrimination result according to the combined weights of the feature combinations in the spatial dimension and the spectral dimension through the discrimination model.

[0014] In a third aspect, the present application provides a computing device, including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a method for selecting structured features of authentic medicinal materials based on Raman peak intensity information as described in the first aspect above.

[0015] In a fourth aspect, the present application provides a computer storage medium storing a computer program, and when the computer program is executed by a computer, it implements a method for selecting structured features of authentic medicinal materials based on Raman peak intensity information as described in the first aspect.

[0016] In this application, a scanning device is used to perform spatial sampling on the surface area of the medicinal material slices to generate two-dimensional distribution data containing concentration gradients of different chemical components. The excitation light generated by a dual-wavelength laser source is alternately projected onto the same sampling area, and a spatial response contrast map is generated based on the Raman peak intensity differences obtained at the two wavelengths. A three-dimensional feature stack is formed according to the spatial response contrast map and the two-dimensional distribution data. Spatial topological analysis is performed on the three-dimensional feature stack to identify at least three target concentration regions and their boundary intersection points. A spatial association model between the target concentration regions is constructed, and the distribution density and direction vector relationship of the boundary intersection points in each target concentration region are analyzed through the spatial association model to generate graph structure data representing the spatial arrangement pattern of the medicinal material components. A structured feature selection operation is performed, and feature combinations that match the prior knowledge of the spatial distribution of genuine medicinal materials are retained in the graph structure data based on the constraint conditions. The selected feature combinations are used as structured features and input into the genuineness discrimination model, so that the discrimination model outputs the discrimination result according to the combined weight of the feature combinations in the spatial dimension and the spectral dimension.

[0017] The present application has the following beneficial effects: Through the scanning device, high-precision spatial sampling is performed on the surface of the medicinal material slices to obtain two-dimensional distribution data of the concentration gradients of chemical components, providing multi-dimensional basic data support for subsequent analysis and ensuring the complete capture of the spatial heterogeneity of chemical components. By alternately projecting the excitation light of the dual-wavelength laser source, a spatial response contrast map of the Raman peak intensity differences is generated, and combined with the two-dimensional distribution data to form a three-dimensional feature stack, enhancing the sensitivity to the spatial-spectral co-variation of chemical components and improving the feature resolution ability. Through topological analysis, the target concentration regions and their boundary intersection points are accurately located, revealing the macroscopic distribution law of the internal chemical components of the medicinal material and providing key node data for spatial association modeling. Analyze the distribution density and direction vector relationship of the boundary intersection points, construct graph structure data representing the component arrangement pattern, transform the complex spatial relationship into a computable topological network, and realize the digital expression of the medicinal material structure. Through the constraint conditions, feature combinations that match the prior knowledge of genuine medicinal materials are screened, redundant information is removed, and high-discriminative features in the spatial-spectral joint dimension are retained, optimizing the input quality of the discrimination model.

[0018] Furthermore, in this application, the characteristic distribution law is extracted from the historical sample library to construct a constraint rule set, and a screening mechanism combining the product value of the regional area dispersion and the intersection point density is used to realize the hierarchical optimization of the feature combination. A dual-channel calculation path is established in the discrimination model, and the joint weight is generated by the asymmetric superposition of the orthogonal grid space fluctuation coefficient and the angle correction factor. The weight spatial transfer is realized through the impedance matching of the annular connection network. Finally, the discrimination mechanism is triggered based on the geometric relationship between the azimuth offset map and the target concentration region.

[0019] Through dynamic constraint rules and multi-dimensional feature quantization, this application accurately preserves the spatial distribution characteristics representing regional authenticity; combined with the spatial transfer mechanism of dual-channel feature fusion and circular network, it enhances the model's ability to analyze the spatial arrangement pattern of medicinal material components, and realizes highly robust regional authenticity identification through the phase comparison of the azimuth offset map, significantly improving the reliability and interpretability of the identification results.

[0020] These aspects or other aspects of this application will be more clearly understood in the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of this application. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0022] Figure 1 Shows a flowchart of a method for selecting structured features of authentic medicinal materials based on Raman peak intensity information provided by this application; Figure 2 Shows a schematic structural diagram of a system for selecting structured features of authentic medicinal materials based on Raman peak intensity information provided by this application; Figure 3 Shows a schematic structural diagram of a computing device provided by this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0023] In order to enable those skilled in the art to better understand the solution of this application, the following will clearly and completely describe the technical solutions in the embodiments of this application with reference to the drawings in the embodiments of this application.

[0024] In some processes described in the specification, claims and above-mentioned drawings of this application, there are multiple operations that appear in a specific order. However, it should be clearly understood that these operations may not be executed in the order in which they appear in this article or may be executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions such as "first" and "second" in this article are used to distinguish different messages, devices, modules, etc., and do not represent a sequence, nor do they limit that "first" and "second" are of different types.

[0025] In the current field of authentic medicinal material identification, traditional Raman spectroscopy technology is generally adopted. Its core defects are reflected in two aspects. On the one hand, existing methods mostly focus on the static analysis of single-point spectral information, and can only obtain qualitative data of local chemical components, but cannot reveal the spatial heterogeneity distribution law of chemical components inside the medicinal materials, resulting in the lack of effective characterization of the unique gradient concentration arrangement pattern of authentic medicinal materials. On the other hand, traditional feature selection strategies overly rely on the screening of single indicators in the spectral dimension, ignoring the synergistic correlation between spatial distribution features and spectral responses, resulting in the loss of key discriminant information. In addition, existing technologies lack the ability to quantitatively analyze the topological structure and its spatial correlation of multi-concentration regions in medicinal material slices, making it difficult to construct an interpretable distribution model. Eventually, the identification model is vulnerable to non-specific feature interference, and its accuracy and robustness are limited.

[0026] To address the above technical bottlenecks, the present invention proposes a structured feature selection method for authentic medicinal materials based on Raman peak intensity information. This method constructs a three-dimensional feature stack through spatial sampling and dual-wavelength excitation technology, fusing multi-dimensional information of two-dimensional distribution data and spatial response contrast maps, breaking through the traditional single spectral analysis dimension; further combining spatial topology analysis and graph structure modeling, accurately extracting the boundary intersection point distribution density and direction vector relationship of the target concentration region, and quantitatively characterizing the spatial arrangement pattern of medicinal material components; finally, through a feature screening mechanism constrained by prior knowledge, retaining the feature combinations that match the spatial structure-activity law of authentic medicinal materials, and realizing highly specific identification based on a spatial-spectral joint weight model. Compared with traditional methods, this solution realizes the collaborative analysis of the spatial heterogeneity distribution of chemical components and spectral responses for the first time through three-dimensional feature fusion and spatial correlation modeling, effectively overcoming the problem of ignoring gradient distribution characteristics in single-point analysis; at the same time, the structured feature selection mechanism significantly improves the extraction efficiency of key discriminant features, reduces noise interference through joint weight calculation, and finally improves the accuracy of authentic medicinal material identification and endows the model with the ability to verify the spatial structure-activity relationship of medicinal materials.

[0027] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative efforts belong to the scope of protection of the present application.

[0028] Figure 1 The following is a flowchart of a structured feature selection method for authentic medicinal materials based on Raman peak intensity information provided for the embodiments of the present application, as Figure 1 shown. This method includes: 101. Perform spatial sampling on the surface area of the medicinal material slice through a scanning device to generate two-dimensional distribution data containing different chemical component concentration gradients.

[0029] Optionally, the spatial sampling of the surface area of the medicinal material slices by the scanning device in step 101 to generate two-dimensional distribution data containing concentration gradients of different chemical components may specifically include: 1011. Irradiate the surface area point by point along a preset grid path through the scanning device, and each grid node corresponds to a sampling position; 1012. At each sampling position, use monochromatic excitation light to vertically irradiate the surface of the medicinal material, synchronously receive the Raman scattering signal reflected at the sampling position, and extract the intensity value within a preset wavelength range in the Raman scattering signal as the characteristic quantity of the grid node; 1013. According to the spatial coordinate arrangement of the grid nodes, fill the characteristic quantities of all grid nodes into a two-dimensional matrix in row-column order, and calculate the difference between the characteristic quantities of adjacent grid nodes. If the difference exceeds the set threshold, insert interpolation nodes at the corresponding row-column gaps to supplement and form a continuously distributed two-dimensional data plane through the characteristic quantities of the interpolation nodes; 1014. Map the characteristic quantity of each sampling position in the two-dimensional data plane to a gray value to generate two-dimensional distribution data with spatial coordinates as the horizontal and vertical axes and gray values representing the concentration of chemical components.

[0030] In the above solution, the scanning device refers to a device that detects the surface of an object point by point through optical or electronic technology and can collect physical or chemical signals related to spatial positions. The grid node refers to the intersection position in the preset grid path, and each node corresponds to a sampling point on the surface of the medicinal material slice, which is used to locate the position of Raman signal collection. The monochromatic excitation light refers to an incident light source with a single wavelength, which is used to excite the surface of the medicinal material to generate Raman scattering signals, and its wavelength selection matches the Raman characteristic peak of the target chemical component. The Raman scattering signal refers to the inelastic scattering light signal generated by the molecules on the surface of the medicinal material under the excitation of monochromatic light, and its wavelength shift is related to the molecular vibration mode and can be used for chemical component analysis. The characteristic quantity refers to the intensity value within a preset wavelength range extracted from the Raman signal, which characterizes the relative concentration of the chemical component corresponding to the sampling position. The interpolation node refers to a virtual node inserted at the gap between the original grid nodes, and its characteristic quantity is calculated through an interpolation algorithm, which is used to fill the mutation region of the concentration gradient between adjacent nodes. The two-dimensional data plane refers to a continuous two-dimensional matrix composed of the characteristic quantities of the grid nodes and the interpolation nodes, which reflects the spatial distribution of the chemical component concentration. The two-dimensional matrix refers to a matrix in which the row index and column index respectively correspond to the physical positions in the horizontal and vertical directions of the surface of the medicinal material.

[0031] In the embodiments of the present application, first, through step 1011, the scanning device is controlled to perform systematic spatial sampling on the surface area according to a preset fine grid path, for example, on the surface of a medicinal material section of 1 cm x 1 cm, on a grid with a step size of 100 μm. The scanning device, such as a confocal Raman microscope, will drive the sample stage or the laser focus to accurately move to the sampling position corresponding to each grid node in turn. For example, starting from the coordinate (0, 0), it scans row by row to (10 mm, 10 mm), generating a total of 100 x 100 sampling points, laying the foundation for subsequent point-to-point measurements.

[0032] Secondly, through step 1012, at each sampling position, the laser source emits monochromatic excitation light, for example, vertically incident on the surface of the medicinal material with a wavelength of 532 nm. The Raman scattering signal reflected is synchronously received by a spectrometer and a CCD detector. Signal processing uses a spectral analysis algorithm, such as fast Fourier transform, to extract the intensity value within a preset wavelength range of 500 - 600 nm. This intensity value is stored as a characteristic quantity. The calculation of the characteristic quantity is achieved by integrating the signal intensity within the preset wavelength range. For example, from the coordinate list obtained in step 1011, each position is processed in turn. After laser irradiation, the detector captures the signal, and the software extracts the intensity value and associates it with the corresponding grid node. This step outputs the characteristic quantity data set for each grid node.

[0033] Then, through step 1013, according to the spatial coordinate arrangement of the grid nodes, the characteristic quantities output in step 1012 are filled into a two-dimensional matrix in row-column order. The difference calculation uses a simple subtraction algorithm to compare the characteristic quantities of adjacent nodes, where the simple subtraction algorithm is for the upper, lower, left, and right neighbors. If the absolute value of the difference exceeds the set threshold, for example, the set threshold = 10, then at the corresponding gap, for example, between rows or columns, interpolation nodes are inserted. Interpolation uses a bilinear interpolation algorithm. According to the characteristic quantities and coordinates of adjacent grid nodes, the characteristic quantity of the new node is calculated, and the formula is (where, is a constant term, and are the linear coefficients of and respectively, is and 's interaction term coefficient). All inserted interpolation nodes and the original nodes together form an extended regular grid, for example, extended from 100 x 100 to 199 x 199, and finally a continuously distributed two-dimensional data plane is generated. For example, first fill the matrix, traverse all adjacent pairs, insert nodes when the difference is too large, and update the matrix.

[0034] Finally, through step 1014, the characteristic quantities of the two-dimensional data plane (including original and interpolation nodes) output in step 1013 are normalized to the range of 0 - 255, using a linear mapping algorithm where feature is the feature quantity of the current node, and are the global minimum maximum value) and converted into grayscale values. With the spatial coordinates as the horizontal axis and the vertical axis , a grayscale image is generated through an image processing library (such as OpenCV). For example, after inputting the feature quantity matrix, the software automatically calculates the global extreme values and completes the grayscale mapping, and finally outputs a two-dimensional distribution image file.

[0035] Taking the detection of saponin components in ginseng slices as an example, the grid spacing is 0.05 mm, the excitation light wavelength is 532 nm, and the threshold value is 2000.

[0036] Specifically, the ginseng slice is fixed on the stage, and the scanning device generates a grid at a spacing of 0.05 mm to cover the surface of the slice. The 785 nm laser is used for excitation, the Raman signal of each node is collected, and the intensity value of the saponin characteristic peak is extracted. It is found that the intensity difference between adjacent nodes in a certain area exceeds the threshold value, and interpolation nodes are inserted for smooth transition. A grayscale distribution map of saponin concentration is generated, showing the gradient characteristic that the saponin in the slice decreases from the center to the outside.

[0037] In this step, through point-by-point scanning and Raman signal extraction, the high-resolution spatial distribution detection of the chemical components on the surface of the medicinal material is realized; through the judgment of the difference threshold and the insertion of interpolation nodes, the continuity and visualization effect of the concentration gradient are enhanced; the finally generated grayscale distribution data plane can intuitively reflect the enrichment area and diffusion trend of the chemical components, providing a reliable basis for the quality evaluation of the medicinal material.

[0038] 102. Alternately project the excitation light generated by the dual-wavelength laser source onto the same sampling area, generate a spatial response contrast map based on the Raman peak intensity differences obtained under the two wavelengths, and form a three-dimensional feature stack according to the spatial response contrast map and the two-dimensional distribution data.

[0039] Optionally, the step of alternately projecting the excitation light generated by the dual-wavelength laser source onto the same sampling area, generating a spatial response contrast map based on the Raman peak intensity differences obtained under the two wavelengths, and forming a three-dimensional feature stack according to the spatial response contrast map and the two-dimensional distribution data in step 102 may specifically include: 1021. Alternately switch the output wavelengths of the dual-wavelength laser source through a synchronous controller, and perform complete spectral acquisition on the same sampling area at each wavelength; 1022. Mark the intensity value of the Raman signal characteristic peak collected at the first wavelength as the first response value, and mark the intensity value of the Raman signal characteristic peak collected at the second wavelength as the second response value; 1023. Calculate the difference between the first response value and the second response value for each spatial sampling point, and map the difference into the same spatial coordinate system as the two-dimensional distribution data to generate a spatial response comparison map; 1024. Use the spatial response comparison map as an additional signal layer and stack it with the two-dimensional distribution data to form a three-dimensional feature stack with three signal intensities.

[0040] In the above solution, the dual-wavelength laser source refers to a laser device that can output two fixed wavelengths, such as 532 nm and 785 nm, and alternately switches the wavelengths to excite the differential responses of target chemical components. The synchronization controller refers to an electronic module used to coordinate the laser wavelength switching and the spectral acquisition timing, ensuring that the Raman signals at the two wavelengths are accurately acquired in chronological order. The Raman peak intensity difference refers to the difference in the intensities of the Raman characteristic peaks generated by the same chemical component under excitation at different wavelengths, reflecting the influence of wavelength selectivity on the detection sensitivity. The spatial response comparison map refers to a two-dimensional matrix with spatial coordinates as the reference and difference intensities as the values, used to visualize the response differences of chemical components under excitation at different wavelengths. The three-dimensional feature stack refers to a three-dimensional data structure formed by stacking the two-dimensional distribution data and the spatial response comparison map layer by layer, containing the original concentration distribution and wavelength response difference information.

[0041] In the embodiments of the present application, first, the synchronization controller sends a timing pulse signal through step 1021 to drive the dual-wavelength laser source to alternately switch between two preset wavelengths, and the switching frequency is usually on the order of kHz to ensure sampling synchronization. At each wavelength, a confocal Raman microscope is used to perform point-by-point spectral acquisition on the same sampling area. During the acquisition process, a spectrometer is used to record the full-band Raman spectra of each spatial point, and the scanning path is controlled by software to ensure coverage of the entire two-dimensional sampling area. Algorithmically, spatial registration techniques are used to ensure that the sampling point coordinates are consistent at different wavelengths. The output is two complete spectral data sets.

[0042] Secondly, through step 1022, the Raman characteristic peak intensity of each sampling point at the first wavelength is marked as the first response value, and the corresponding peak intensity at the second wavelength is marked as the second response value. For example, if the characteristic peak intensity of a certain sampling point is 8000 photons under 532 nm excitation and 5000 photons under 785 nm, they are respectively recorded as the first and second response values I1 and I2.

[0043] Then, for each sampling point output by Step 1022, the first wavelength response value I1 and the second wavelength response value I2 are used to perform a dual-wavelength response difference calculation for each position. The absolute difference algorithm ΔI = |I1 - I2| is used to quantify the response difference. For example, at the sampling point (2, 3), I1 = 850 and I2 = 420, then ΔI = |850 - 420| = 430. Next, strictly inherit the spatial coordinate arrangement established in Step 101, including the exact coordinates of the original grid nodes and all interpolation nodes, to create an empty matrix M_ΔI that is exactly the same size as the two-dimensional distribution data. For example, a 5×5 grid corresponds to a 5×5 matrix. Traverse each sampling point according to the coordinate index, and accurately fill the ΔI value of this point into the corresponding position of M_ΔI. For example, the ΔI = 430 at the coordinate (2, 3) is filled into M_ΔI[2][3]. For the interpolation nodes inserted in Step 101, synchronously calculate and fill their ΔI values to ensure full spatial coverage. Finally, through the matrix-to-image conversion algorithm, the M_ΔI matrix is mapped into a spatial response contrast map, where the horizontal / vertical axis corresponds to the spatial coordinates, and the pixel gray value represents the ΔI intensity. For example, ΔI = 0 is displayed as black, and ΔI ≥ 500 is displayed as white, generating a spatial response contrast map that is strictly aligned with the original concentration distribution.

[0044] Finally, in Step 1024, the spatial response contrast map is denoted as matrix M_ΔI, with the same size of m×n. Each element M_ΔI[i][j] represents the dual-wavelength response difference value at the same position, and the two-dimensional distribution data obtained in Step 101 is denoted as matrix M_gray, with the size of m×n. Each element M_gray[i][j] represents the chemical component concentration gray value at the spatial coordinate (i, j). Then, create a three-dimensional array data structure with the dimension of m×n×2. Take M_gray as the first signal layer at Z = 0 layer and completely copy it to the D[:,:,0] channel of the three-dimensional array, that is, for each spatial position (i, j), assign D[i][j][0] = M_gray[i][j]. Then, take M_ΔI as the second signal layer at Z = 1 layer and copy it to the D[:,:,1] channel, that is, for each position (i, j), assign D[i][j][1] = M_ΔI[i][j]. Finally, if there is an expansion requirement, such as adding a third signal layer, other feature data can be continuously stacked at the Z = 2 layer to form a complete three-dimensional feature stacked data structure. For example, for a 3×3 grid system, the original concentration matrix is [[180, 200, 190], [210, 220, 200], [190, 180, 210]], and the response contrast matrix is [[310, 290, 300], [280, 270, 290], [290, 310, 280]]. Then, the three-dimensional feature at the position (1, 1) after stacking is [220, 270].

[0045] Taking the detection of ferulic acid in Angelica sinensis slices as an example, specifically, a dual-wavelength laser source alternately irradiates the same area of the Angelica sinensis slice, and Raman signals are collected respectively. To extract the intensity of the ferulic acid characteristic peak, the recorded value at 532 nm is the first response value, and the value at 830 nm is the second response value. When calculating the difference, it is found that the response value at 532 nm in the epidermal region is 2000 photons higher than that at 830 nm, while it is 1500 photons lower in the medulla region. The generated spatial response contrast map shows that the epidermis is a high-difference region. Superimposing the contrast map with the ferulic acid concentration distribution map, the three-dimensional stack shows that the epidermal region has both high concentration and high wavelength response differences simultaneously.

[0046] This step enhances the ability to identify the spatial response characteristics of specific chemical components through dual-wavelength excitation and difference comparison; the three-dimensional feature stack integrates the concentration distribution and wavelength selectivity information, which can reveal the response law of chemical components under different excitation conditions, provides a data basis for multimodal analysis, and at the same time avoids the limitations of single-wavelength detection.

[0047] 103. Perform spatial topological analysis on the three-dimensional feature stack to identify at least three target concentration regions and their boundary intersection points.

[0048] Optionally, the performing spatial topological analysis on the three-dimensional feature stack in step 103 to identify at least three target concentration regions and their boundary intersection points may specifically include: 1031. Perform a segmentation operation on each signal layer in the three-dimensional feature stack, and extract the continuous region with a signal intensity value higher than the preset threshold as the candidate concentration region; 1032. Compare the spatial positions of the candidate concentration regions in multiple signal layers, and retain the candidate regions with a spatial overlap degree exceeding the set ratio in at least two signal layers as the target concentration regions; 1033. Detect the boundary curves of each target concentration region, and calculate the minimum distance and the curvature change rate between the boundary curves corresponding to adjacent target concentration regions; 1034. When the minimum distance between the boundary curves corresponding to adjacent target concentration regions is less than the preset ratio of the regional average width, and the curvature change rate exceeds the preset mutation threshold, it is determined that there is an intersection point between the boundaries corresponding to adjacent target concentration regions.

[0049] In the above solution, three-dimensional feature stacking refers to a three-dimensional data structure formed by stacking multiple layers of signal data, such as medical images, remote sensing data, or sensor network data, along the spatial dimension. Each layer contains the spatial distribution of signal intensity values. The candidate concentration region refers to a continuous region extracted by a segmentation operation in a single layer of signal data, whose signal intensity value is higher than a preset threshold, such as the gray threshold of a lesion region in a medical image. The spatial overlap degree refers to the proportion of the overlapping area of candidate regions in different signal layers projected onto the same spatial coordinate system, which is used to measure the consistency of cross-layer regions. The boundary curve refers to the outer contour of the target concentration region, which is a continuous curve fitted by discrete points and characterizes the shape of the region. The regional average width refers to the average span of the target concentration region in the main extension direction, which is used to normalize the determination criterion of the boundary spacing. The curvature change rate refers to the derivative of the local curvature of the boundary curve, which reflects the degree of mutation of the boundary shape, such as sharp corners or inflection points. The boundary intersection point is determined by detecting that the gradient change rate of adjacent regions exceeds a set change amplitude.

[0050] In the embodiment of the present application, first, each signal layer of the three-dimensional feature stacking is independently segmented through step 1031. The Otsu algorithm is used to automatically calculate the optimal threshold for the first signal layer, traversing the gray levels from 0 to 255, and selecting the threshold T0 that maximizes the between-class variance σ² = ω0ω1(μ0 - μ1)². After generating a binary mask, 8-neighborhood connected component labeling is performed, and the set C0 of candidate concentration regions is output. Synchronously, a specific fixed threshold T1 for the medicinal material is set for the second signal layer. After denoising by morphological opening operation with a 3×3 structuring element, the connected regions are labeled, and the set C1 of candidate concentration regions is output. For example, the ferulic acid enrichment region R1 is segmented from the concentration layer of angelica slices, and the dual-wavelength response region S1 is segmented from the difference layer.

[0051] Secondly, through step 1032, according to the spatial coordinate arrangement and the spatial response comparison diagram, it is ensured that the regions C0 and C1 are in the same spatial reference system. An affine transformation matrix is used to correct the position offset for spatial position comparison, where the formula is is the coordinate before transformation, is the coordinate after transformation, is the rotation and scaling matrix, is the translation vector; the Jaccard overlap degree of each pair of regions is calculated, and the formula is , where is the number of overlapping pixels, is the number of pixels in the union. Only the region combinations with an overlap degree > 0.6 are retained in the calculation result as the set T of target concentration regions. For example, if the overlap degree between R1 and S1 is 0.75, the target region T1 is generated.

[0052] Then, through step 1033, boundary extraction is performed on each target region. The Moore boundary tracing algorithm is used to start from the top-left pixel and search the 8-neighborhood in the counterclockwise direction to generate a closed boundary curve For adjacent target regions (e.g., and ), calculate the Hausdorff distance between the boundary curves (where are the boundary point coordinates, represents the point to the curve of the shortest Euclidean distance), and calculate the average curvature change rate (where is the tangent angle of the -th boundary point, is the total number of boundary points). For example, the spacing between the astragaloside region and the polysaccharide region , the curvature change rate radians .

[0053] Finally, first calculate the average width of the target region through step 1034, and use the volume equivalent formula for the target region to obtain (where represents the average width, and the volume is the total volume of the target region). Set the judgment condition. If the minimum spacing is less than and the curvature change rate exceeds the mutation threshold , then mark the boundary crossing. In this example and , so at the three-dimensional coordinates is confirmed as the boundary crossing point (where is the minimum spacing, is the curvature change rate, represents the rate of change of curvature along the path).

[0054] Taking the 5×5 mm section of the genuine medicinal material A as an example, specifically, process the pixel concentration layer and the difference layer of the three-dimensional feature stack respectively: The concentration layer uses Otsu adaptive threshold segmentation (threshold ), extract the region , and its coordinate range is (10 - 30, 20 - 40); the difference layer is segmented by a fixed threshold , and the region is obtained, with the coordinate range of (15 - 32, 22 - 42); calculate the overlap degree of the two regions, and generate the target region ; for adjacent target regions , calculate the boundary parameters through step 1033: the minimum Hausdorff distance ; the average curvature change rate ; radian ; Judgment condition: spacing condition ; Curvature condition ; Finally, in the three-dimensional coordinates Identify as a boundary intersection point.

[0055] This step achieves accurate identification of the three-dimensional target area through layer-by-layer segmentation and spatial registration, and effectively locates the intersection based on the boundary geometric features. Its advantage is that it combines multi-level signal correlation to avoid the limitations of single-layer analysis, and at the same time improves the robustness of boundary topology analysis through curvature changes and spacing criteria, which is suitable for automated analysis of complex three-dimensional structures.

[0056] 104. Construct a spatial correlation model between the target concentration areas, and analyze the distribution density and directional vector relationship of the boundary intersection points in each target concentration area through the spatial correlation model to generate graph structure data representing the spatial arrangement pattern of medicinal material components.

[0057] Optionally, the step 104 of analyzing the distribution density and directional vector relationship of the boundary intersection points in each target concentration area by the spatial correlation model to generate graph structure data representing the spatial arrangement pattern of the medicinal material components may specifically include: 1041. Establishing an initial topological network with the geometric center of each target concentration area as a node and the boundary intersection as a connection hub through the spatial association model; 1042. In the initial topological network, the corresponding number of intersections within a preset radius around each boundary intersection is calculated as a density distribution parameter, and the direction angle of the line connecting each boundary intersection and the center of gravity of the target concentration area to which it belongs is recorded as a direction relationship parameter; 1043. Perform vector synthesis on boundary intersections with overlapping density distribution parameters to generate weighted edges representing the strength of interaction between regions; 1044. Use the density distribution parameter and the direction relationship parameter as node attributes, use the vector synthesis direction of the weighted edge as an edge attribute, and construct graph structure data containing multi-level spatial association features.

[0058] In the above scheme, the target concentration area refers to the distribution area of medicinal material component concentration divided by image analysis or chemical detection. Each area represents a high concentration cluster of specific chemical components, including geometric shape, boundary, concentration gradient and other attributes. The boundary intersection point refers to the intersection point of the boundaries of different target concentration areas, which is used to characterize the spatial contact relationship between areas. The geometric center of gravity refers to the geometric center position of the target concentration area, which is obtained by calculating the average coordinates of the polygon vertices of the regional outline and serves as the node of the topological network. The initial topological network refers to an undirected graph with the geometric center of gravity as the node and the boundary intersection point as the connection hub, which describes the adjacency relationship between regions. The density distribution parameter refers to the number of other intersection points within a preset radius with the boundary intersection point as the center, reflecting the density of the regional boundary. The direction relationship parameter refers to the direction angle of the line connecting the boundary intersection point and the geometric center of gravity of the area to which it belongs, which is used to quantify the contribution direction of the intersection point to the area. The weighted edge refers to the edge generated by vector synthesis, whose weight represents the strength of interaction between regions, and the direction is determined by the angle of the synthetic vector. The graph structure data refers to a topological network containing node attributes such as density and direction angle, and edge attributes such as weight and direction, which is used to describe the spatial arrangement pattern of medicinal material components.

[0059] In the embodiment of the present application, first, step 1041 is used to calculate the geometric center of gravity coordinates of each target concentration area set according to the target concentration area set and the boundary intersection point set output in step 103, wherein the formula for calculating the center of gravity coordinates is: ,in is the geometric centroid coordinate of the kth target concentration area, n is the total number of boundary intersections contained in the area, and (xi,yi) is the spatial coordinate of the ith boundary intersection. With the centroid as the node and the intersection as the hub, an initial topological network is established, where the node set V={ }, hub set H = { }, edge set ,in Indicates area For example, the centroids of the three saponin-rich regions G1-G3 and the boundary intersections P1-P5 in Astragalus slices were used to construct a network containing 3 nodes and 5 hubs.

[0060] Next, at step 1042, the density distribution parameter of each boundary intersection is calculated. ,by Draw a circle with a preset radius r as the center, for example, r=0.5mm, and count the number of other intersections in the circle .in For intersection The density value at are the coordinates of other boundary intersection points. Synchronously calculate the direction relationship parameter θm and obtain the vector The direction angle of , , where is the centroid of the region to which the set of boundary intersection points belongs. Calculate the centroid of the region and the direction angle of each boundary intersection point according to the above steps, and record the direction angle of the line connecting the centroid of the region to which each boundary intersection point belongs and the centroid of the target concentration region as the direction relationship parameter.

[0061] Subsequently, identify points with the same or similar density distribution parameters through step 1043, and perform vector synthesis on the identified intersection points with overlapping density distributions >0. Extract the direction vectors of all boundary intersection point sets in the same region. The formula is , and weighted synthesize the region representative vector , where the weight , is the sum of all in the same region. Generate a weighted edge e ij between the geometric centroid nodes of different target concentration regions, and the weight . The weight w ij of the weighted edge e ij directly quantifies the interaction strength between target concentration regions to generate a weighted edge representing the interaction strength between regions. For example, the included angle between the synthesis vectors of the saponin region and the polysaccharide region is 60°, and an edge with a weight of 0.87 is generated.

[0062] Finally, through step 1044, for each target concentration region, summarize all the boundary intersection points associated with it (calculate the arithmetic mean of their density distribution parameters as the node density attribute to quantify the complexity of the boundary structure of the region, indicating the dense distribution of multiple intersection points; at the same time, extract the region representative vector generated in step 1043, and use its direction angle as the node direction attribute to characterize the overall distribution trend of chemical components in space. For each weighted edge, use the included angle between the two region representative vectors calculated in step 1043 as the edge direction attribute, and directly inherit the weight as the edge strength attribute. The two jointly quantify the spatial interaction mode. Finally, integrate the three-layer association features to construct graph structure data, where the first-layer region feature is the node attribute describing the structure properties of the target concentration region itself; the second-layer interaction feature is the edge attribute capturing the spatial interaction relationship between adjacent regions; the third-layer topological feature is the network connection structure revealing the spatial arrangement framework of multiple regions, forming graph structure data of a multi-level spatial association model.

[0063] Taking the sliced Scutellaria baicalensis Georgi as an example, analyze the spatial distribution of flavonoids in the sliced Scutellaria baicalensis Georgi.

[0064] Specifically, five flavonoid high-concentration regions are divided through microscopic imaging and threshold segmentation, the barycentric coordinates are calculated, 12 boundary intersection points are detected, and a topological network containing 5 nodes and 12 hubs is constructed. A radius of 100 μm is set, 2 to 5 adjacent points around each intersection point are counted, and the direction angles are recorded. For example, the direction angle between intersection point A and the barycenter of region 1 is 45°. For the intersection point group with a density of 3, vectors with direction angles of 30°, 90°, and 150° are synthesized, and a weighted edge with a weight of 2.6 and a direction of 60° is generated. Finally, graph data containing 5 nodes and 8 weighted edges is generated for identifying the "radiation-cluster" distribution pattern of flavonoid components.

[0065] In this step, a multi-level spatial association model is constructed by quantifying the density and direction features of boundary intersection points, which can intuitively reveal the spatial interaction patterns of medicinal material components, improve the analysis efficiency of the arrangement rules of complex components, and provide structured input for the component prediction based on graph neural networks.

[0066] 105. Perform a structured feature selection operation to retain, through constraint conditions, the feature combinations in the graph structure data that match the prior knowledge of the spatial distribution of authentic medicinal materials.

[0067] Optionally, the retaining, through constraint conditions, the feature combinations in the graph structure data that match the prior knowledge of the spatial distribution of authentic medicinal materials in step 105 may specifically include: 1051. Extract the feature distribution law of the graph structure data from the historical authentic medicinal material sample library, and construct a constraint rule set through the feature distribution law; 1052. Traverse all feature combinations in the graph structure data, and screen out the feature combinations that meet the preset conditions as the primary screening set; among them, the feature combinations that meet the preset conditions refer to the feature combinations with multiple constraint marks; 1053. For each feature combination in the primary screening set, calculate the product value of the regional area dispersion and the intersection point density, and retain the feature combinations whose product value ranks within the predefined ratio range as the screened feature combinations.

[0068] In the above solution, the graph structure data refers to a feature network represented in the form of a graph, where nodes represent feature attributes, such as the climate and soil attributes of the distribution area of medicinal materials, and edges represent the association relationships between features. The feature combination refers to at least two parameters among the number of boundary intersection points, the area ratio of the target concentration region, and the included angle of the direction vector. The feature distribution law refers to the spatial distribution pattern of feature combinations in historical genuine medicinal material samples, such as the correlation between high altitude areas and specific precipitation. The constraint rule set includes a set of logical rules such as the threshold of regional area dispersion, the interval of intersection point density, and the distribution pattern of vector included angles, such as "genuine medicinal materials need to simultaneously satisfy altitude > 1000 meters and precipitation > 800 mm". The constraint mark refers to an identifier indicating that a feature combination satisfies a certain constraint rule, such as being marked as rule A or rule B. The regional area dispersion refers to the degree of dispersion of the spatial area distribution corresponding to the feature combination, such as the standard deviation or entropy value, which reflects the size of the regional area difference. The intersection point density refers to the ratio of the number of topological intersection points corresponding to the feature combination to the regional area, which reflects the complexity of the boundary.

[0069] In the embodiment of the present application, first, extract the feature distribution law of the graph structure data generated in step 1044 from the historical genuine medicinal material sample library through step 1051; perform Gaussian mixture model clustering on the density mean of the node attributes, the reverse angle, and the included angle and weight of the edge attributes of all samples in the historical genuine medicinal material sample library to form a feature matrix, for example, the covariance matrix type is full. Calculate the probability distribution of each feature matrix. For each feature matrix, take the arithmetic mean of the eigenvalues as the confidence interval to generate constraint rules. Merge feature matrices with strong correlations to generate a constraint rule set.

[0070] Subsequently, through step 1052, traverse all possible feature combinations in the graph structure data of the current medicinal material. Each feature combination includes at least the density mean or direction angle of one node attribute and the included angle α or weight of the associated edge attribute, such as the attribute of node G1 in the saponin area and the attribute α12 = 42° of the edge e12 connecting it to the polysaccharide area form a combination c1; then use the constraint rule set to perform rule matching on each combination, and screen out feature combinations that match two rules simultaneously. Among them, initialize the counter count = 0. If each combination satisfies rule 1, then count + 1, and if it satisfies rule 2, then count + 1. For example, in c1 ∈[2.8, 3.4] triggers rule 1, and α12 = 42° < 50° triggers rule 2, resulting in count = 2. Only when count ≥ 2, mark each combination as a valid combination; finally, integrate all combinations with a mark number ≥ 2, such as c1. Construct a primary screening set. For example, 15 double-rule matching combinations are screened out in the astragalus slices.

[0071] Finally, for each feature combination in the primary screening set, calculate its area dispersion degree through step 1053. The formula is , where δ k is the area dispersion degree of the feature combination, is the actually measured area of a certain feature combination, is the predefined standard reference area. Let and the actually measured obtain and the intersection point density , where obtain , calculate the product value and retain the top 15% of the combinations before sorting, and finally output the screened feature combinations that match the prior knowledge of the spatial distribution of genuine regional drugs.

[0072] Take the Chinese herbal medicine Astragalus membranaceus as an example.

[0073] Specifically, discover the feature combination rules of the genuine producing area of Astragalus membranaceus from historical data, such as "annual average temperature 6 - 10°C, soil pH 6.5 - 7.5, slope < 15°". Then screen out the feature combinations that simultaneously meet the temperature, pH, and slope rules, and eliminate the combinations that only meet a single condition. And calculate the area dispersion degree and intersection point density of the retained combinations, and retain the combinations with low dispersion degree and high density.

[0074] This step realizes the precise optimization of feature combinations through historical law constraints and geometric feature screening. Its advantage lies in integrating prior knowledge to enhance feature correlation, while combining regional distribution stability and topological complexity to effectively distinguish key features from redundant features, and enhancing the interpretability and reliability of the spatial distribution model of genuine regional drugs.

[0075] 106. Input the screened feature combinations as structured features into the authenticity discrimination model, so as to output the discrimination result according to the joint weights of the feature combinations in the spatial dimension and the spectral dimension through the discrimination model.

[0076] Optionally, the step of inputting the screened feature combinations as structured features into the authenticity discrimination model in step 106 to output the discrimination result according to the joint weights of the feature combinations in the spatial dimension and the spectral dimension through the discrimination model may specifically include: 1061. Input the screened feature combinations as discrete units with spatial coordinate markings into the authenticity discrimination model; 1062. Establish a dual-channel calculation path inside the authenticity discrimination model. The first channel maps the two-dimensional coordinate values of each discrete unit to a virtual plane divided by an orthogonal grid, calculates the standard deviation of the spectral response amplitude differences of all discrete units within each orthogonal grid, and forms a spatial fluctuation coefficient in units of grids. The second channel extracts the straight-line distance values of the intersection points of each discrete unit with the boundary of its target concentration region, and generates an angle correction factor in combination with the directional correlation intensity. 1063. Generate a combined weight value through an asymmetric superposition method based on the spatial fluctuation coefficient and the angle correction factor. 1064. Construct a calculation network containing a ring connection structure, inject the combined weight value of each discrete unit into its corresponding ring node, realize the spatial transfer of the weight value through the impedance matching relationship between adjacent ring nodes, enable the output end of the calculation network to receive the combined weight values output from each port, and perform a phase comparison between the sum of the combined weight values of each port and a preset azimuth reference value to generate a phase comparison result. 1065. Generate an azimuth offset map based on the phase comparison result, perform an overlapping analysis on the spatial distribution patterns of the extreme offset regions that continuously appear in the azimuth offset map and the target concentration region. When a specific angular relationship is formed between the boundary of the extreme offset region and the geometric centers of at least three target concentration regions, trigger the confirmation mechanism of the authenticity discrimination model and output the discrimination result.

[0077] In the above solution, the discrete unit refers to the combination of features after screening. Each unit contains spatial coordinates such as two-dimensional longitude and latitude or image pixel positions, and corresponding spectral feature data such as spectral response values in different bands. The orthogonal grid refers to a mathematical partitioning method that divides a virtual plane into a grid of rows and columns with equal spacing. The grid boundaries are composed of straight lines that intersect perpendicularly, and are used to statistically analyze spatial distribution characteristics. The spatial fluctuation coefficient refers to the standard deviation calculated based on the amplitude difference of the spectral responses of discrete units within the orthogonal grid, which reflects the stability or difference of spectral features in the region. The angle correction factor refers to a parameter jointly generated by the straight-line distance from the discrete unit to the boundary of the target concentration region and the directional correlation strength, and is used to correct the directional deviation in weight calculation. The combined weight value refers to the comprehensive weight obtained by asymmetrically superimposing the spatial fluctuation coefficient and the angle correction factor, which characterizes the importance of the feature combination in the spatial and spectral dimensions. The ring connection structure refers to a network topology structure. Each node corresponds to a discrete unit, and its adjacent nodes are connected through a ring path. The weight value is transmitted between nodes through impedance matching. The azimuth offset map refers to a two-dimensional map generated based on phase comparison, which shows the distribution of offsets in the region. The extreme value region represents the region with abnormal spatial distribution or significant features. The impedance matching relationship means that the impedance value is inversely proportional to the cosine component of the directional correlation strength between units. Phase comparison means that at the output end of the calculation network, a receiving array with direction selectivity is set. The array contains acquisition ports evenly distributed at equal angles along eight azimuths. Each port only receives the weight value transmitted by the ring node from a specific azimuth angle, and compares the cumulative weight value of each port with a preset azimuth reference value in terms of phase.

[0078] In the embodiment of the present application, first, through step 1061, the screened feature combination, such as the node attributes and associated edge attributes of the saponin region that satisfy the spatial distribution rule in the astragalus slicing, is combined. The spatial position information it carries, that is, the surface coordinates of the medicinal material corresponding to this feature combination, such as the two-dimensional grid position x = 0.2 mm, y = 0.3 mm recorded by Raman scanning, is marked as the inherent attribute of the discrete unit and input into the authenticity discrimination model.

[0079] Secondly, through step 1062, a dual-channel calculation path is established. The first channel maps the discrete unit coordinates to the virtual plane divided by the orthogonal grid. For the discrete units within each grid, the standard deviation of the amplitude difference of their spectral responses is calculated. The formula is The spatial fluctuation coefficient is obtained, where is the standard deviation of the amplitude difference of the spectral responses, is the th spectral response amplitude of the discrete unit, is the mean value of the spectral response amplitudes of all discrete units within the grid, is the total number of discrete units within the grid. The second channel measures the straight-line distance from each discrete unit to the nearest boundary demarcation point of its affiliated area. For example, the distance from a saponin area unit to the demarcation point is 0.05 mm. Combining the boundary correlation weight of this demarcation point and the angle between the direction from the discrete unit to the demarcation point and the main component distribution direction, an angle correction factor is obtained through direction consistency adjustment.

[0080] Next, the spatial fluctuation coefficient is obtained through step 1063 as the stability reference value, and (1 + λ) is used as the multiplication coefficient, where λ is the angle correction factor. The two calculation results are multiplied to obtain the final combined weight.

[0081] Subsequently, through step 1064, a circular calculation network is constructed, mapping each discrete unit. For example, the coordinates of a saponin area unit (0.2 mm, 0.3 mm) are mapped to a circular node, and the initial weight value of the node is the combined weight value calculated in step 1063. Subsequently, the dynamic impedance value between adjacent nodes is calculated, and the formula for impedance is , w A and w B are the weight values of nodes A and B. For example, the weight of node A is 1.54 and the weight of adjacent node B is 1.30, then , and the weight is converted into a transmission signal through the current transmission model. For example, node A outputs current amperes to the adjacent node; when the network runs, the signal converges from the boundary nodes to the center, and finally the total current values of each port are accumulated at the output end. For example, the currents of the three output ends are [2.1∠30°, 1.8∠45°, 3.4∠20°]. The vector phase of this combined current vector is compared with the reference value of the genuine regional medicinal materials orientation, and the formula for the phase difference is , where arg() is the complex vector phase angle function. For example, the phase difference ΔΦ = 15° between the actual combined phase of 30° and the reference phase of 15°, generating a phase comparison result.

[0082] Finally, through step 1065, the data in the phase comparison result, such as the deviation ΔΦ = 14° between the network output phase of 32° and the reference phase of 18°, is used to bind the spatial coordinates of each discrete unit with the phase difference to generate an azimuth offset map. The phase offset values of all units are marked in the coordinate system. For example, the area where ΔΦ > 10° is rendered as a red heat zone to form a two-dimensional offset matrix containing spatial distribution information. Subsequently, spatial registration of the azimuth offset map and the target concentration area is performed. The azimuth offset map is superimposed on the two-dimensional distribution data of the medicinal material section, and the continuously occurring offset extreme value areas in the figure are extracted. For example, a red area with an area greater than 0.1 mm² is used as the target to be analyzed. Then, geometric relationship detection is carried out. The boundary contour line of the offset extreme value area is calculated, and the included angle relationship formed with the geometric center points of at least three target concentration areas is measured. Among them, a triangular network is constructed with the center point group, and the included angle θ is formed by calculating the perpendicular feet from the boundary line of the offset area to the connecting lines of each centroid. For example, the included angle between the perpendicular foot from the boundary point to the AB connecting line is 118°, the included angle between the perpendicular foot to the BC connecting line is 122°, and the included angle between the perpendicular foot to the AC connecting line is 120°. Finally, when the included angle relationship matches a specific genuine regional structure rule, such as 115° ≤ θ ≤ 125° and the extreme value area covers the midpoint of the connecting line of the centroids, the confirmation mechanism of the genuine regional discrimination model is triggered. For example, since the astragalus membranaceus section simultaneously satisfies the triangular included angle of 120° ± 2° and the coverage rate of the offset area reaches 85%, the identification result of the genuine medicinal material is output.

[0083] Taking the identification of the genuineness of the origin of Chinese medicinal materials as an example, a 10×10 pixel grid, with a spatial fluctuation coefficient weight of 0.6 and an angle correction factor weight of 0.4.

[0084] Specifically, for the hyperspectral image data of Chinese medicinal materials in a certain area, each pixel point contains spatial coordinates and spectral characteristics. The pixel points are used as discrete units and input into the model, and the coordinates correspond to the image positions. An orthogonal grid of 0×10 pixels is divided, and the standard deviation of the spectral response amplitude within each grid is calculated; at the same time, the distance and direction angle from the pixel to the boundary of the preset genuine production area are calculated. The spatial fluctuation coefficient with a weight of 0.6 and the angle correction factor with a weight of 0.4 are superimposed to generate a combined weight. A circular network is constructed to transfer the weight, and the port corresponding to the larger phase difference is output as the abnormal area. It is found that the offset extreme value area forms an equilateral triangle included angle with the centers of three genuine production areas, and it is determined as a genuine medicinal material.

[0085] This step combines spatial and spectral information through a dual-channel calculation path, combines the spatial transfer characteristics of the circular network, dynamically adjusts the combined weight value, and finally realizes highly robust genuineness identification through phase comparison and morphological overlap analysis. Its advantage lies in being able to adapt to the characteristic distribution differences in different regions, reduce noise interference, and verify the reliability of the results through geometric relationships, significantly improving the accuracy and interpretability of the identification.

[0086] The following is a complete embodiment of steps 101 to 106: In a system for authenticating genuine medicinal materials based on spatial-spectral fusion, first, 10 samples of genuine medicinal materials from Region A and 10 samples of ordinary medicinal materials from Region B were selected. Transverse sections with a thickness of 50 μm were prepared using a precision slicing machine and placed on a microscopic confocal Raman scanning platform for spatial sampling. The 5×5 mm² area was scanned with a step precision of 0.5 μm, and the original number of sampling points was downsampled to a 500×500 grid during actual processing. Each pixel represents a 10 μm×10 μm area to generate a two-dimensional chemical composition distribution matrix.

[0087] Secondly, a dual-wavelength laser module was configured with wavelengths of 830 nm and 785 nm. The two excitation lights were alternately projected onto the same sampling area through an optical fiber coupler. When the 830 nm laser was detected, the system recorded the intensity of the characteristic Raman peak at 1600 cm⁻¹; when switched to 785 nm, the intensity of the characteristic peak at 1250 cm⁻¹ was captured. A spatial response comparison map was generated through difference calculation, and the pixel intensity values of this map were used as the third dimension (Z3) and registered and stacked with the previous two-dimensional distribution data to form a 256×256×3 three-dimensional feature stack containing spatial coordinates + three characteristic values.

[0088] Then, a density clustering algorithm was used to perform spatial topology analysis on the three-dimensional stack. By setting the constraint conditions of flavonoid concentration threshold > 2.5 mg / g, saponin concentration gradient change rate < 15%, and Raman response difference > 300, target concentration regions α (diameter 1.2 mm), β (diameter 0.8 mm), and γ (diameter 0.5 mm) were identified in the northwest quadrant of the section. The Voronoi diagram partitioning algorithm was used to determine the boundary intersection points of the three regions, and a total of 5 characteristic intersection sites were co-located.

[0089] Next, a spatial association model based on graph theory was constructed. Each intersection point was used as a node, and the Euclidean distance between nodes was calculated as the edge weight. At the same time, the characteristic that the node distribution density in the northwest quadrant reached 18 per mm² was statistically analyzed. Through direction vector analysis, it was found that the nodes in region α showed a radial distribution at a 30° angle, and the nodes in region β had a circular arrangement with a 60° interval. These topological features were quantified as an adjacency matrix and stored as graph structure data.

[0090] Subsequently, according to the "Spatial Distribution Standard of Genuine Medicinal Materials", the characteristic screening rules were set: ① node density threshold ≥ 15 per mm²; ② radial angle deviation < 5°; ③ circular arrangement integrity > 80%. Through these three constraint conditions, 12 characteristic combinations that met the requirements were screened out from the 28 characteristics of the original graph structure data, including key indicators such as spatial distribution dispersion (0.32) and characteristic angle matching degree.

[0091] Finally, the filtered structured features are input into a pre-trained authenticity discrimination model. The authenticity model determines that the matching degree of the sample with the authentic medicinal materials in Area A reaches 87.3% in terms of the spatial arrangement pattern through the combined calculation of the spatial dimension weight of 0.6 and the spectral dimension weight of 0.4, while the matching degree with the common medicinal materials in Area B is only 41.2%. Finally, it outputs the identification conclusion that the sample conforms to the characteristics of authentic medicinal materials.

[0092] This embodiment fully demonstrates the technical process from data collection to result interpretation. All technical parameters are virtually set and do not involve specific brand and regional information, meeting the confidentiality requirements. Each step includes specific operation methods, data processing processes, and decision-making logics, forming a closed-loop technical verification system.

[0093] Figure 2 The following is a schematic structural diagram of a system for selecting structured features of authentic medicinal materials based on Raman peak intensity information provided by an embodiment of the present application. As Figure 2 shown, the system includes: A first generation module 21, configured to perform spatial sampling on the surface area of a medicinal material slice through a scanning device to generate two-dimensional distribution data containing different chemical component concentration gradients; A second generation module 22, configured to alternately project the excitation light generated by a dual-wavelength laser source onto the same sampling area, generate a spatial response comparison map based on the Raman peak intensity differences obtained under two wavelengths, and form a three-dimensional feature stack according to the spatial response comparison map and the two-dimensional distribution data; An identification module 23, configured to perform spatial topology analysis on the three-dimensional feature stack to identify at least three target concentration regions and their boundary intersection points; A third generation module 24, configured to construct a spatial association model between the target concentration regions, and analyze the distribution density and direction vector relationship of the boundary intersection points in each target concentration region through the spatial association model to generate graph structure data representing the spatial arrangement pattern of the medicinal material components; An output module 25, configured to perform structured feature selection operations, retain the feature combinations that match the prior knowledge of the spatial distribution of authentic medicinal materials in the graph structure data based on constraint conditions, and use the filtered feature combinations as structured features to be input into the authenticity discrimination model, so as to output an identification result according to the combined weights of the feature combinations in the spatial dimension and the spectral dimension through the discrimination model.

[0094] Figure 2 The system for selecting structured features of authentic medicinal materials based on Raman peak intensity information can execute Figure 1For the method for selecting the structured features of genuine medicinal materials based on Raman peak intensity information described in the foregoing embodiments, its implementation principle and technical effects will not be elaborated herein. For the system for selecting the structured features of genuine medicinal materials based on Raman peak intensity information in the above embodiments, the specific manners in which each module and unit perform operations have been described in detail in the embodiments related to the method, and will not be elaborated herein.

[0095] In a possible design, Figure 2 the system for selecting the structured features of genuine medicinal materials based on Raman peak intensity information in the foregoing embodiments can be implemented as a computing device, such as Figure 3 shown, the computing device may include a storage component 31 and a processing component 32; The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32.

[0096] The processing component 32 is used for the above Figure 2 method for selecting the structured features of genuine medicinal materials based on Raman peak intensity information in the foregoing embodiments.

[0097] Among them, the processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented by one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components for executing the above method.

[0098] The storage component 31 is configured to store various types of data to support the operation of the terminal. The storage component may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0099] Of course, the computing device may also necessarily include other components, such as an input / output interface, a display component, a communication component, etc.

[0100] The input / output interface provides an interface between the processing component and the peripheral interface module, and the above peripheral interface module may be an output device, an input device, etc.

[0101] The communication component is configured to facilitate wired or wireless communication between the computing device and other devices, etc.

[0102] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform. In this case, the computing device can refer to a cloud server. The above processing components, storage components, etc. can be basic server resources leased or purchased from a cloud computing platform.

[0103] The embodiment of the present application also provides a computer storage medium storing a computer program, and when the computer program is executed by a computer, it can implement the above Figure 1 structured feature selection method of authentic Chinese medicinal materials based on Raman peak intensity information shown in the embodiment.

[0104] Those skilled in the art can clearly understand that for the convenience and conciseness of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated herein.

[0105] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0106] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0107] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present application.

Claims

1. A method for selecting the structured features of authentic Chinese medicinal materials based on Raman peak intensity information, characterized in that include: The surface area of the medicinal material slice is spatially sampled by a scanning device to generate two-dimensional distribution data containing concentration gradients of different chemical components; The excitation light generated by the dual-wavelength laser source is alternately projected onto the same sampling area, a spatial response comparison graph is generated based on the difference in Raman peak intensity obtained at the two wavelengths, and a three-dimensional feature stack is formed based on the spatial response comparison graph and the two-dimensional distribution data; performing spatial topological analysis on the three-dimensional feature stack to identify at least three target concentration regions and their boundary intersections; Constructing a spatial correlation model between the target concentration regions, and analyzing the distribution density and directional vector relationship of the boundary intersection points in each target concentration region through the spatial correlation model to generate graph structure data representing the spatial arrangement pattern of the medicinal material components; A structured feature selection operation is performed to retain, through constraints, feature combinations that match the prior knowledge of the spatial distribution of authentic medicinal materials in the graph structure data, and the screened feature combinations are input as structured features into the authenticity discrimination model, so as to output the identification results through the discrimination model according to the joint weights of the feature combinations in the spatial dimension and the spectral dimension.

2. The method according to claim 1, wherein The method of retaining a feature combination matching the prior knowledge of spatial distribution of authentic medicinal materials in the graph structure data through constraint conditions includes: Extracting characteristic distribution rules of the graph structure data from a historical authentic medicinal material sample library, and constructing a constraint rule set based on the characteristic distribution rules; Traversing all feature combinations in the graph structure data, and screening out feature combinations that meet preset conditions as primary screening sets; wherein the feature combinations that meet preset conditions refer to feature combinations with multiple constraint tags; For each feature combination in the primary screening set, the product value of the regional area dispersion and the intersection density is calculated, and the feature combinations whose product values are ranked within a predefined ratio range are retained as the screened feature combinations.

3. The method according to claim 1, wherein The screened feature combination is input as a structured feature into the authenticity discrimination model, so that the discrimination model outputs the identification result according to the joint weight of the feature combination in the spatial dimension and the spectral dimension, including: The filtered feature combination is input into the authenticity discrimination model as a discrete unit with spatial coordinates; A dual-channel calculation path is established inside the authenticity discrimination model, in which the first channel maps the two-dimensional coordinate value of each discrete unit to a virtual plane divided by an orthogonal grid, calculates the standard deviation of the spectral response amplitude difference of all discrete units in each orthogonal grid, and forms a spatial fluctuation coefficient in grid units; the second channel extracts the straight-line distance value between each discrete unit and the boundary intersection of the target concentration area to which it belongs, and generates an angle correction factor in combination with the directional correlation strength; Generate a joint weight value according to the spatial fluctuation coefficient and the angle correction factor by an asymmetric superposition method; Construct a computing network including a ring connection structure, inject the joint weight value of each discrete unit into its corresponding ring node, realize the spatial transmission of the weight value through the impedance matching relationship between adjacent ring nodes, enable the output end of the computing network to receive the joint weight value output by each port, and compare the sum of the joint weight values of each port with the preset azimuth reference value to generate a phase comparison result; An azimuth offset map is generated according to the phase comparison result, and the spatial distribution forms of the offset extreme value areas and the target concentration areas that appear continuously in the azimuth offset map are overlapped and analyzed. When the boundary of the offset extreme value area forms a specific angle relationship with the geometric center of at least three target concentration areas, the confirmation mechanism of the authenticity discrimination model is triggered, and the identification result is output.

4. The method according to claim 1, characterized in that The spatial correlation model is used to analyze the distribution density and directional vector relationship of the boundary intersection points in each target concentration area to generate graph structure data representing the spatial arrangement pattern of medicinal material components, including: An initial topological network is established through the spatial association model, with the geometric center of gravity of each target concentration area as a node and the boundary intersection as a connection hub; In the initial topological network, the corresponding number of intersections within a preset radius around each boundary intersection is calculated as a density distribution parameter, and the direction angle of the line connecting each boundary intersection and the center of gravity of the target concentration area to which it belongs is recorded as a direction relationship parameter; Vector synthesis of boundary intersections with overlapping density distribution parameters is performed to generate weighted edges representing the strength of interactions between regions; The density distribution parameter and the direction relationship parameter are used as node attributes, and the vector synthesis direction of the weighted edge is used as an edge attribute to construct graph structure data containing multi-level spatial association features.

5. The method according to claim 1, wherein The surface area of the medicinal material slice is spatially sampled by a scanning device to generate two-dimensional distribution data containing concentration gradients of different chemical components, including: The surface area is irradiated point by point along a preset grid path by a scanning device, and each grid node corresponds to a sampling position; At each sampling position, a monochromatic excitation light is vertically incident on the surface of the medicinal material, a Raman scattering signal reflected at the sampling position is synchronously received, and an intensity value of a preset wavelength range in the Raman scattering signal is extracted as a feature value of the grid node; According to the spatial coordinate arrangement of the grid nodes, the feature quantities of all the grid nodes are filled into the two-dimensional matrix in row-column order, and the difference between the feature quantities of adjacent grid nodes is calculated. If the difference exceeds the set threshold, an interpolation node is inserted in the corresponding row-column gap to supplement the feature quantities of the interpolation nodes to form a continuously distributed two-dimensional data plane; The characteristic quantity of each sampling position in the two-dimensional data plane is mapped into a grayscale value to generate two-dimensional distribution data with spatial coordinates as horizontal and vertical axes and grayscale values representing chemical component concentrations.

6. The method according to claim 1, wherein The method alternately projects the excitation light generated by the dual-wavelength laser source onto the same sampling area, generates a spatial response comparison graph based on the difference in Raman peak intensity obtained at the two wavelengths, and forms a three-dimensional feature stack based on the spatial response comparison graph and the two-dimensional distribution data, including: The output wavelengths of the dual-wavelength laser source are alternately switched by a synchronization controller, and complete spectral acquisitions are performed on the same sampling area at each wavelength; The Raman signal characteristic peak intensity values collected at the first wavelength are marked as the first response values, and the Raman signal characteristic peak intensity values collected at the second wavelength are marked as the second response values; For each spatial sampling point, calculate the difference between the first response value and the second response value, and map the difference to the same spatial coordinate system as the two-dimensional distribution data to generate a spatial response comparison map; Take the spatial response comparison map as an additional signal layer and stack it with the two-dimensional distribution data to form a three-dimensional feature stack with three signal intensities.

7. The method according to claim 1, characterized in that, Performing spatial topology analysis on the three-dimensional feature stack to identify at least three target concentration regions and their boundary intersection points, including: Performing a segmentation operation on each signal layer in the three-dimensional feature stack, and extracting continuous regions with signal intensity values higher than a preset threshold as candidate concentration regions; Perform a spatial position comparison on the candidate concentration regions in multiple signal layers, and retain the candidate regions with a spatial overlap degree exceeding a set ratio in at least two signal layers as target concentration regions; Detect the boundary curves of each target concentration region, and calculate the minimum distance and the curvature change rate between the boundary curves corresponding to adjacent target concentration regions; When the minimum distance between the boundary curves corresponding to adjacent target concentration regions is less than a preset ratio of the regional average width and the curvature change rate exceeds a preset mutation threshold, it is determined that there is an intersection point between the boundaries corresponding to adjacent target concentration regions.

8. A structured feature selection system for genuine medicinal materials based on Raman peak intensity information, characterized in that, Including: A first generation module for spatially sampling the surface area of the medicinal material slice through a scanning device to generate two-dimensional distribution data containing different chemical composition concentration gradients; A second generation module for alternately projecting the excitation light generated by the dual-wavelength laser source onto the same sampling area, generating a spatial response comparison map based on the Raman peak intensity differences obtained at the two wavelengths, and forming a three-dimensional feature stack according to the spatial response comparison map and the two-dimensional distribution data; An identification module for performing spatial topology analysis on the three-dimensional feature stack to identify at least three target concentration regions and their boundary intersection points; A third generation module for constructing a spatial association model between the target concentration regions, and analyzing the distribution density and direction vector relationship of the boundary intersection points in each target concentration region through the spatial association model to generate graph structure data representing the spatial arrangement pattern of the medicinal material components; An output module for performing a structured feature selection operation, retaining a feature combination matching the prior knowledge of the spatial distribution of authentic medicinal materials in the graph structure data based on a constraint condition, and using the selected feature combination as structured features to input into an authenticity discrimination model, so as to output a discrimination result according to the combined weights of the feature combination in the spatial dimension and the spectral dimension through the discrimination model.

9. A computing device, characterized in that, Including a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement the method for selecting structured features of authentic medicinal materials based on Raman peak intensity information according to any one of claims 1 to 7.

10. A computer storage medium, characterized in that, A computer program is stored, and when the computer program is executed by a computer, it implements the method for selecting the structured features of genuine medicinal materials based on Raman peak intensity information as described in any one of claims 1 to 7.

Images