Mural pigment detection method and device based on multispectral analysis

Abstract

The invention provides a mural pigment detection method and device based on multispectral analysis, and belongs to the technical field of spectral analysis, and the method comprises the following steps: obtaining first actual spectral data and second actual spectral data for representing mural components; extracting first spectral feature data from the first actual spectral data, and extracting second spectral feature data from the second actual spectral data; and according to the first spectral feature data and the second spectral feature data, determining the pigment composition of the mural. According to the method, the color composition of the mural is comprehensively and accurately detected by combining different types of spectral data and utilizing collaborative analysis of multi-dimensional spectral feature information.

Description

Technical Field

[0001] The present invention relates to the technical field of spectral analysis, and in particular to a mural pigment detection method and device based on multi-spectral analysis. Background Art

[0002] Due to their age, ancient murals have been significantly impacted by both natural and human factors, resulting in various defects such as pigment loss and discoloration. Therefore, with the continuous development of cultural heritage protection, mural restoration, and materials science research, the demand for pigment composition analysis technology is growing.

[0003] Traditional pigment analysis methods primarily include chemical reagent analysis and microscopic observation. While these methods can provide certain analytical results, they often require sampling or processing of the murals, potentially causing irreversible damage. Furthermore, simple laser-induced breakdown spectroscopy (LIBS) can only obtain elemental information, while simple Raman spectroscopy can only obtain molecular information. Furthermore, most detection methods are limited to surface testing and cannot comprehensively and accurately determine the color composition of the murals from all angles.

[0004] Therefore, how to comprehensively and accurately detect the color composition of murals has become a technical problem that needs to be solved urgently. Summary of the Invention

[0005] The present invention provides a mural pigment detection method and device based on multi-spectral analysis, which is used to overcome the defect of the existing technology that it is limited to surface detection, and realizes comprehensive and accurate detection of the color composition of the mural.

[0006] The present invention provides a method for detecting mural pigments based on multispectral analysis, comprising the following steps: Acquiring first actual spectrum data and second actual spectrum data for characterizing components of the mural; extracting first spectrum characteristic data from the first actual spectrum data, and extracting second spectrum characteristic data from the second actual spectrum data; The pigment composition of the mural is determined according to the first spectral characteristic data and the second spectral characteristic data.

[0007] According to a mural pigment detection method based on multispectral analysis provided by the present invention, extracting first spectral characteristic data from the first actual spectral data and extracting second spectral characteristic data from the second actual spectral data include: Calculating a first-order derivative and a second-order derivative of each first data point in the first actual spectrum data; determining at least one first characteristic peak in the first actual spectrum data according to changes in the first-order derivative and the second-order derivative of each of the first data points; Obtaining the first spectral characteristic data according to all the first characteristic peaks; Calculating a first-order derivative and a second-order derivative of each second data point in the second actual spectrum data; determining at least one second characteristic peak in the second actual spectrum data according to changes in the first-order derivative and the second-order derivative of each second data point; The second spectrum characteristic data is obtained based on all the second characteristic peaks.

[0008] According to a mural pigment detection method based on multispectral analysis provided by the present invention, determining the pigment composition of the mural according to the first spectral characteristic data and the second spectral characteristic data includes: According to the first spectral feature data and the second actual spectral data, respectively obtaining a first spectral posterior probability of the first spectral feature data in each category and a second spectral posterior probability of the second spectral feature data in each category; Performing Bayesian fusion on the first spectral posterior probability corresponding to each category and the second spectral posterior probability corresponding to each category to calculate the fused posterior probability of each category; The pigment composition of the mural is determined based on all the fused posterior probabilities.

[0009] According to a mural pigment detection method based on multispectral analysis provided by the present invention, obtaining, based on the first spectral feature data and the second actual spectral data, a first spectral posterior probability of the first spectral feature data in each category and a second spectral posterior probability of the second spectral feature data in each category, respectively, includes: Acquire first sample spectral data and second sample spectral data; Calculating a first priori probability of the first sample spectral data in each category, and calculating a second priori probability of the second sample spectral data in each category; Calculating a first mean and a first variance of the first sample spectral data and the first spectral feature data respectively, and calculating a second mean and a second variance of the second sample spectral data and the second spectral feature data respectively; Calculating a first likelihood probability of each first data point in the first spectral feature data under each category according to the first mean and the first variance; Calculating a second likelihood probability of each second data point in the second spectral feature data under each category according to the second mean and the second variance; Calculating a first spectral posterior probability corresponding to each category based on the first prior probability corresponding to each category and the first likelihood probability corresponding to each category; The second spectral posterior probability corresponding to each category is calculated according to the second prior probability corresponding to each category and the second likelihood probability corresponding to each category.

[0010] According to a mural pigment detection method based on multispectral analysis provided by the present invention, determining the pigment composition of the mural according to all the fused posterior probabilities includes: determining a maximum fused posterior probability from all of the fused posterior probabilities; The pigment composition of the mural is determined according to the pigment category corresponding to the maximum fused posterior probability.

[0011] According to a mural pigment detection method based on multispectral analysis provided by the present invention, determining the pigment composition of the mural according to the first spectral characteristic data and the second spectral characteristic data further includes: determining all element types of the mural according to the first spectral characteristic data; determining all molecular structure information of the mural according to all the element types and the second spectral characteristic data; The pigment composition of the mural is determined based on all the element types and all the molecular structure information.

[0012] According to a mural pigment detection method based on multispectral analysis provided by the present invention, determining all element types of the mural based on the first spectral characteristic data includes: Matching the first spectral characteristic data with all first preset characteristic information in a preset standard element database one by one; When the first spectral characteristic data matches the first preset characteristic information, all element types of the mural are determined according to the element types corresponding to each matched first preset characteristic information.

[0013] According to a mural pigment detection method based on multispectral analysis provided by the present invention, determining all molecular structure information of the mural based on all the element types and the second spectral characteristic data includes: Determining, based on all the element types, second preset characteristic information corresponding to all the element types in a preset molecular component database; each second preset characteristic information is associated with one element type; Matching the second spectral characteristic data with all the second preset characteristic information one by one; When the second spectral characteristic data matches all of the second preset characteristic information, all of the molecular structure information of the mural is determined according to the molecular structure information corresponding to each piece of the second preset characteristic information.

[0014] According to a mural pigment detection method based on multispectral analysis provided by the present invention, before obtaining the first actual spectral data and the second actual spectral data for characterizing the mural components, the method further includes: When the pulsed laser emits light, the laser-induced breakdown spectrum data is collected by a first spectrometer, and when the continuous laser emits light, the Raman spectrum data is collected by a second spectrometer; Normalization processing, smoothing and denoising processing, and spectral baseline correction processing are performed on the laser-induced breakdown spectroscopy data and the Raman spectroscopy data, respectively, to obtain the first actual spectral data and the second actual spectral data, respectively.

[0015] The present invention also provides a mural pigment detection device based on multi-spectral analysis, comprising: Housing, switch button, display screen, grip, battery, data interface, trigger and light outlet; The switch button is provided on the outer surface of the housing; the display screen is embedded in the outer surface of the housing; the upper end of the handle is fixedly connected to the housing; the battery is detachably mounted on the lower end of the handle; the data interface is provided on the outer surface of the housing; the trigger is provided at the front end of the handle; the light exit hole is provided at the front end of the housing; The mural pigment detection device based on multispectral analysis also includes: a first controller, a second controller, a first spectrometer, a continuous laser, a second spectrometer, a Raman probe, a Raman fiber connection end, a continuous laser fiber connection end, an N20 reduction motor, a pulse laser, a lead screw, a ball guide rail, a LIBS light output hole, a Raman light receiving hole, and a LIBS light receiving hole; The first controller, the second controller, the first spectrometer, the continuous laser, the second spectrometer, the Raman probe, the Raman fiber connection end, the continuous laser fiber connection end, the N20 reduction motor, the pulse laser, the lead screw, the ball guide rail, the LIBS light output hole, the Raman light transceiver hole, and the LIBS light receiving hole are all fixedly installed inside the housing; The first controller is electrically connected to the display screen and is in communication with the second controller; The second controller is electrically connected to the continuous laser and the pulse laser respectively to control the light output of the continuous laser and the pulse laser respectively; The continuous laser is connected to the Raman probe via the continuous laser optical fiber connection end; The Raman probe is arranged on the ball guide rail and moves forward and backward along the screw rod under the drive of the N20 reduction motor; The Raman light transceiver hole is provided at the front end of the Raman probe, and the Raman probe is used to transmit continuous laser light to the sample surface through the Raman light transceiver hole and receive Raman light signals; The LIBS light exit hole is provided at the front end of the pulse laser, and the pulse laser is used to emit LIBS light to the sample surface through the LIBS light exit hole, and receive the LIBS light signal after passing through the sample surface through the LIBS light receiving hole; The first spectrometer and the second spectrometer are respectively connected to the Raman fiber connection end and the LIBS light receiving hole through a Z-type optical fiber; wherein the Z-type optical fiber includes a branching structure, which splits the LIBS optical signal into two paths, one of which directly enters the second spectrometer, and the other is combined with the Raman optical signal and then enters the first spectrometer; The output ends of the first spectrometer and the second spectrometer are both electrically connected to the first controller for transmitting spectral data to the display screen for display.

[0016] In summary, one or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: By acquiring first and second actual spectral data used to characterize the composition of the mural, comprehensive multidimensional spectral information of the sample is obtained. By extracting first spectral signature data from the first actual spectral data and second spectral signature data from the second actual spectral data, more discriminative characteristic peak information is obtained, respectively, eliminating invalid information and background interference from the spectral data. Furthermore, by determining the pigment composition of the mural based on the first and second spectral signature data, the collaborative analysis of multidimensional spectral signature information enables comprehensive and accurate detection of the mural's color composition. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 The figure is a flow chart of the mural pigment detection method based on multi-spectral analysis provided by the present invention.

[0019] Figure 2This is one of the external structural schematic diagrams of the mural pigment detection device based on multi-spectral analysis provided by the present invention.

[0020] Figure 3 This is the second schematic diagram of the external structure of the mural pigment detection device based on multi-spectral analysis provided by the present invention.

[0021] Figure 4 This is one of the internal structure diagrams of the mural pigment detection device based on multi-spectral analysis provided by the present invention.

[0022] Figure 5 This is the second schematic diagram of the internal structure of the mural pigment detection device based on multi-spectral analysis provided by the present invention.

[0023] Figure 6 This is the third schematic diagram of the internal structure of the mural pigment detection device based on multi-spectral analysis provided by the present invention.

[0024] Figure 7 The mural pigment detection device provided by the present invention is used for atomic emission spectrometry detection of realgar and orpiment pigments.

[0025] Figure 8 The invention discloses a Raman spectrum of a mural pigment detection device for detecting realgar and orpiment pigments.

[0026] Figure numerals: 0, housing; 1, switch button; 2, display screen; 3, grip; 4, battery; 5, data interface; 6, trigger; 7, light output hole; 8, first controller; 9, second controller; 10, first spectrometer; 11, continuous laser; 12, second spectrometer; 13, Raman probe; 14, Raman fiber connection end; 15, continuous laser fiber connection end; 16, N20 reduction motor; 17, pulse laser; 18, lead screw; 19, ball guide rail; 20, LIBS light output hole; 21, Raman light transceiver hole; 22, LIBS light receiving hole. DETAILED DESCRIPTION

[0027] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0028] It should be noted that, in the description of the present invention, the terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the presence of other identical elements in the process, method, article or device comprising the element. The orientation or positional relationship indicated by the terms "upper", "lower", etc. is based on the orientation or positional relationship shown in the accompanying drawings, and is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the system or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to the specific circumstances.

[0029] The terms "first," "second," and so forth, used herein are used to distinguish similar objects, not to describe a specific order or precedence. It should be understood that such terms are interchangeable where appropriate, allowing embodiments of the present invention to be implemented in an order other than that illustrated or described herein. Furthermore, the terms "first," "second," and so forth generally distinguish objects of a single type, and do not limit the number of objects. For example, the first object may be one or more. Furthermore, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates an "or" relationship between the connected objects.

[0030] The following combination Figures 1-8 The present invention describes a mural pigment detection method, system, electronic device and storage medium based on multi-spectral analysis.

[0031] Figure 1 This is one of the flow diagrams of the mural pigment detection method based on multi-spectral analysis provided by the present invention, such as Figure 1 As shown, including but not limited to the following steps: Step 101: Acquire first actual spectral data and second actual spectral data for characterizing components of a mural.

[0032] When implementing step 101, to fully characterize the composition of the mural, it is necessary to separately acquire first and second actual spectral data. Because murals are composed of multiple superimposed pigment layers, they typically exhibit complex multi-layered pigment properties, with pigments distributed across different layers, each potentially composed of different chemical compositions. This complicates detection and analysis. The simultaneous acquisition of these two types of spectral data is necessary because the first spectral data provides elemental information about the mural, while the second spectral data provides information on its molecular structure. The combination of these two types of data effectively overcomes the limited information dimensionality of a single spectral technique, enabling a comprehensive analysis of the mural's pigment composition.

[0033] In a possible implementation, before step 101, the following steps are further included: When the pulsed laser emits light, the laser-induced breakdown spectrum data is collected by the first spectrometer, and when the continuous laser emits light, the Raman spectrum data is collected by the second spectrometer; Normalization processing, smoothing and denoising processing, and spectral baseline correction processing are performed on the laser-induced breakdown spectroscopy data and the Raman spectroscopy data, respectively, to obtain first actual spectral data and second actual spectral data, respectively.

[0034] Specifically, the purpose of acquiring laser-induced breakdown spectroscopy (LIBS) and Raman spectroscopy (Raman spectroscopy) data is to provide a complete spectral information foundation for subsequent analysis. LIBS data uses a pulsed laser to excite the sample surface to generate plasma, which is used to capture elemental information of the mural, while Raman spectroscopy uses a continuous laser to excite the sample's molecular vibrations, which is used to identify its molecular structure information. This design is because elemental information and molecular structure information are determined by different spectral characteristics, respectively. A single laser and a single spectrometer cannot simultaneously complete comprehensive detection, so it is necessary to obtain two types of initial spectral data.

[0035] More specifically, to obtain laser-induced breakdown spectroscopy (LIBS) data, a pulsed laser is controlled to emit high-power pulsed laser light, which is focused onto the mural surface via an optical system to excite the sample and generate a plasma. The emitted light from the plasma passes through a receiving probe and enters a first spectrometer, which decomposes it into spectral signals of varying wavelengths, generating LIBBS data. LIBBS, with its high sensitivity and rapid response, is fundamental to the precise detection of elemental species and their underlying characteristics in samples.

[0036] To obtain Raman spectral data, a continuous laser is controlled to emit a stable continuous laser beam. This laser beam is precisely focused onto the mural surface through an optical system, inducing molecular vibrations in the sample. The Raman scattered light caused by these molecular vibrations is collected by a microscope objective and transmitted through an optical path to a second spectrometer, which decomposes the optical signal to produce Raman spectral data. By capturing molecular vibrational patterns, Raman spectroscopy can clearly reveal the molecular structure of a sample, making it particularly suitable for non-destructive analysis.

[0037] Furthermore, in order to ensure the high quality and accuracy of the laser-induced breakdown spectroscopy data and Raman spectroscopy data, they need to be preprocessed. The preprocessing methods for laser-induced breakdown spectroscopy data and Raman spectroscopy data are the same. In the following text, spectral data will be directly used to refer to laser-induced breakdown spectroscopy data and Raman spectroscopy data. Specifically, preprocessing includes normalization, smoothing and denoising, and spectral baseline correction. The purpose of these operations is to eliminate noise, background interference and dimensional differences in the spectral data, enhance the clarity of the target information, and thus lay a solid foundation for subsequent feature extraction and component analysis.

[0038] First, the spectral data is normalized. This embodiment uses the min-max normalization method to map the numerical range of the spectral data to between 0 and 1, eliminating the impact of the original data dimension and making different data indicators comparable. The normalization formula is: ; Where x is the original data value, is the normalized value, min is the minimum value in the data set, and max is the maximum value in the data set. Normalization ensures that the signal intensities of different spectra are within the same numerical range, providing a consistent data foundation for subsequent analysis.

[0039] Next, the normalized spectral data is smoothed and denoised. This embodiment uses the Savitzky-Golay smoothing method, which is a smoothing technique based on polynomial fitting. It can remove high-frequency noise while retaining the key features of the spectral signal, including the position, shape, and width of the spectral peak. Its smoothing formula is: ; in, is the smoothed value of the i-th data point, is the Savitzky-Golay smoothing coefficient, which is pre-calculated based on the window size 2k+1 and the polynomial order m. is the value of the original data at position i+j. This method is suitable for spectral data with obvious peaks and can effectively remove noise while retaining key spectral features.

[0040] Finally, the smoothed spectral data is baseline corrected. This embodiment uses the least squares baseline correction method (Asymmetric Least Squares, ALS), and its formula is: ; in, is the original spectral data, is the estimated baseline, is the weight, when hour, Larger, otherwise smaller, to ensure that the peak signal is preserved; is the smoothing parameter, which is used to control the smoothness of the baseline. is the second derivative of the baseline, which is used to ensure the smoothness of the baseline. The corrected spectral signal is: ; Baseline correction can effectively eliminate the interference of background noise or baseline drift on spectral data, thereby more accurately analyzing the peak information of the spectral signal.

[0041] Through normalization, smoothing, denoising and baseline correction, the laser-induced breakdown spectroscopy data and Raman spectroscopy data are converted into first actual spectral data and second actual spectral data with high signal-to-noise ratio and background interference removed.

[0042] Step 102: extracting first spectrum characteristic data from the first actual spectrum data, and extracting second spectrum characteristic data from the second actual spectrum data.

[0043] When implementing step 102, in order to accurately determine the element types of the mural, it is necessary to extract at least one first spectral feature data from the first actual spectral data and at least one second spectral feature data from the second actual spectral data, thereby laying the foundation for subsequent spectral analysis.

[0044] In a possible implementation, step 102 specifically includes the following steps: Step 201: Calculate the first-order derivative and the second-order derivative of each first data point in the first actual spectrum data.

[0045] Step 202: Determine at least one first characteristic peak in the first actual spectrum data according to changes in the first-order derivative and the second-order derivative of each first data point.

[0046] Step 203: Obtain first spectrum characteristic data according to all first characteristic peaks.

[0047] Step 204: Calculate the first-order derivative and the second-order derivative of each second data point in the second actual spectrum data.

[0048] Step 205: Determine at least one second characteristic peak in the second actual spectrum data according to changes in the first-order derivative and the second-order derivative of each second data point.

[0049] Step 206: Obtain second spectrum characteristic data based on all second characteristic peaks.

[0050] In steps 201 to 206 above, the calculation of the first-order derivative can reflect the rate of change of the spectral signal. By identifying the position of the zero crossing point, the potential spectral peak area can be preliminarily located. The calculation of the second-order derivative further captures the curvature change of the spectral signal, especially highlighting the position of the characteristic peak. When calculating the first-order and second-order derivatives, the discrete difference method can be used, among which the most commonly used is the central difference method. For the i-th data point, the approximate value of its first-order derivative is: ; The approximate value of the second-order derivative is: ; Where h is the interval between data points, For spectral data at point The strength of the place.

[0051] In practice, by calculating the derivative of each data point in the spectral data, first-order and second-order derivative curves can be generated. The zero crossing points of the first-order derivative curve are generally associated with the location of the characteristic peak, while the negative extreme points of the second-order derivative curve can further confirm the existence and significance of the peak. In this way, the combined effect of the first-order and second-order derivatives can significantly improve the reliability and accuracy of spectral peak identification.

[0052] Through derivative calculation, the slowly changing baseline interference in spectral data can be effectively eliminated, making the characteristic peaks more prominent, while enhancing the ability to capture weak characteristic peaks in complex samples.

[0053] Through the combined action of the first-order derivative and the second-order derivative, the position and characteristics of the characteristic peak can be accurately located. The zero crossing point of the first-order derivative corresponds to the extreme point of the spectral signal, that is, the possible characteristic peak position, and the negative value interval of the second-order derivative can further confirm the authenticity of the characteristic peak and distinguish the characteristic peak from the pseudo-peak or noise signal. In actual operation, for the first-order derivative curve, the position of the zero crossing point is marked as a candidate point of the potential characteristic peak; then, by checking the second-order derivative value of the corresponding point, if it is negative, the position is confirmed to be a characteristic peak. That is, when and For spectral data of multi-layer pigments, this process can effectively identify the significant characteristic peaks of each layer, avoiding missed detections or false detections due to signal overlap or background interference.

[0054] In this way, the characteristic peaks in the spectral signal can be determined one by one, including the specific location and significance of the peaks. This method not only reduces manual intervention but also improves the accuracy and efficiency of characteristic peak identification.

[0055] By determining characteristic peaks based on derivative changes, the reliability and comprehensiveness of characteristic peak identification in the first actual spectral data can be significantly improved. Furthermore, first spectral characteristic data can be obtained based on all first characteristic peaks. This allows for accurate capture of characteristic information for each pigment layer in spectral analysis of complex samples, even when faced with multiple layers of pigments or overlapping signals. This provides high-quality data support for subsequent feature extraction, element identification, and composition analysis.

[0056] Therefore, in the above steps 201 to 206, the first actual spectrum data (i.e., the pre-processed laser induced breakdown spectrum data) and the second actual spectrum data (i.e., the pre-processed Raman spectrum data) are respectively obtained by finding their corresponding characteristic peaks by taking the first-order derivative and the second-order derivative, thereby obtaining the first spectrum characteristic data and the second spectrum characteristic data.

[0057] Step 103: Determine the pigment composition of the mural according to the first spectral characteristic data and the second spectral characteristic data.

[0058] In this embodiment, step 103 aims to determine the pigment composition of the mural based on the first and second spectral signature data. This step, as the core decision-making step in the entire method, aims to leverage the complementary features extracted from the two different spectral data to accurately identify the pigment composition of the target sample, thereby meeting the dual requirements of accurate component identification and multi-source data fusion in the cultural relic analysis process.

[0059] Specifically, in the aforementioned steps, the first actual spectral data and the second actual spectral data are pre-processed and peak-searched respectively, so as to obtain the first spectral characteristic data and the second spectral characteristic data respectively. In step 103, the above two types of spectral characteristic data are combined for analysis to finally determine the pigment composition of the target sample. This analysis process can be implemented through a variety of implementation methods. It can be based on a unified classification model constructed based on a probabilistic statistical method, and the first spectral characteristic data and the second spectral characteristic data are input respectively to obtain respective category prediction results and make fusion decisions; it can also be based on expert rules or a preset matching mechanism, first identifying the constituent elements from the first spectral characteristic data, and then combining the second spectral characteristic data to determine the molecular structure, and finally comprehensively judging the pigment type. Regardless of which specific path is adopted, the core mechanism is: mapping the spectral characteristics derived from different physical processes to a unified pigment classification space, and realizing effective judgment of the target sample category in this space.

[0060] The following examples will illustrate these two methods in detail.

[0061] In a possible implementation, step 103 may be implemented through steps 301 to 303: Step 301: According to the first spectral feature data and the second actual spectral data, respectively obtain a first spectral posterior probability of the first spectral feature data in each category and a second spectral posterior probability of the second spectral feature data in each category.

[0062] In this embodiment, step 301 is used to determine the posterior probabilities of the first and second spectral signatures for each pigment category. This step aims to provide a quantitative basis for subsequent fusion decisions based on probabilistic reasoning, thereby improving the accuracy and stability of pigment classification.

[0063] In practice, the first spectral signature data is extracted from the first actual spectral data (i.e., the preprocessed LIBS spectrum) through operations such as second-order derivative peak finding. It primarily reflects the distribution of elemental components in the sample. The second spectral signature data is extracted from the second actual spectral data (i.e., the preprocessed Raman spectrum). It primarily reflects the vibrational modes of the sample's molecular structure or organic functional groups. Because different mineral pigment categories exhibit certain statistical regularities in elemental composition and molecular structure, a unified Bayesian classification model can be constructed to evaluate these two types of signature data separately and output their posterior probabilities for each candidate category.

[0064] Specifically, the Bayesian classification model is based on the prior probability and likelihood probability, and calculates the posterior probability of each category through Bayes' theorem. In terms of prior probability, the model performs normalization processing based on the number of samples in each category in the training sample to obtain an initial probability estimate of the occurrence of each category. In terms of likelihood probability, the model is based on the spectral feature distribution under each category in the training set, and calculates the fitting probability of the first spectral feature data and the second spectral feature data under this category respectively. Assuming that each spectral feature obeys a Gaussian distribution, the model will score the likelihood of the currently input feature data in each category based on the mean and variance parameters. By multiplying the prior probability with the likelihood probability and normalizing it, the spectral posterior probability of the target spectral feature data in each category can be obtained.

[0065] In a possible implementation, step 301 specifically includes steps 401 to 407: Step 401: Acquire first sample spectral data and second sample spectral data.

[0066] Step 402: Calculate a first priori probability of the first sample spectral data in each category, and calculate a second priori probability of the second sample spectral data in each category.

[0067] Step 403: Calculate the first mean and first variance of the first sample spectral data and the first spectral feature data respectively, and calculate the second mean and second variance of the second sample spectral data and the second spectral feature data respectively.

[0068] Step 404: Calculate the first likelihood probability of each first data point in the first spectral feature data in each category according to the first mean and the first variance.

[0069] Step 405: Calculate the second likelihood probability of each second data point in the second spectral feature data in each category according to the second mean and the second variance.

[0070] Step 406: Calculate the first spectral posterior probability corresponding to each category based on the first prior probability corresponding to each category and the first likelihood probability corresponding to each category.

[0071] Step 407: Calculate the second spectral posterior probability corresponding to each category based on the second prior probability corresponding to each category and the second likelihood probability corresponding to each category.

[0072] In this embodiment, steps 401 through 407 calculate the posterior probabilities of the first and second spectral features for each mineral pigment category based on the statistical and spectral characteristics of the training samples. This process, based on Bayes' theorem and combining prior probabilities with likelihood functions, constructs a comprehensive posterior probability calculation mechanism, which is crucial for improving spectral classification accuracy.

[0073] In step 401, first and second sample spectral data are acquired to train classification models for the first and second actual spectral data (i.e., LIBS spectra) and Raman spectra, respectively. The first and second sample spectral data can be composed of a large number of samples with known labels, each corresponding to a known mineral pigment class. This sample data provides a realistic observational foundation for subsequent statistical modeling.

[0074] In step 402, for each category , and count the first sample number in the first sample spectral data and the second sample number in the second sample spectral data, respectively, and record them as and , and according to the total number of all samples in the first sample spectral data and the total number of all samples in the second sample spectral data , where R represents the sample type (first or second). According to the formula: The first prior probability corresponding to each category can be calculated separately With the second prior probability The prior probability represents the initial probability that a sample belongs to a certain category under unobserved conditions and is one of the basic parameters of Bayesian classification.

[0075] In step 403, for each category in the first sample spectral data and the second sample spectral data , calculate the first mean under each feature dimension respectively and first variance , and the second mean and the second variance These statistical parameters are used to construct a Gaussian distribution model to describe the distribution of spectral features in each category in each dimension, and are the core parameters for subsequent likelihood function calculations.

[0076] In step 404, the first spectral feature data of the sample to be tested is substituted into the Gaussian probability density function under each category, the likelihood probability of each feature dimension is calculated respectively, and the first likelihood probability of each first data point in the first spectral feature data under each category is obtained according to the following formula: : in, Represents the value of the j-th first data point in the first spectral feature data.

[0077] Similarly, in step 405, in the same manner, according to the second mean and the second variance , calculate the second likelihood probability of each second data point in the second spectral feature data under each category The specific calculation method will not be described in detail.

[0078] In step 406, first, the first likelihood probability of each first data point in the first spectral feature data under each category is calculated. , calculate the overall first spectral feature data in each category The first joint likelihood probability under : Furthermore, according to Bayes' theorem, the first prior probability corresponding to each category is combined with the first joint likelihood probability corresponding to each category to calculate the first spectral posterior probability corresponding to each category. The specific calculation formula is: In step 407, the same Bayesian posterior inference method as in step 406 is used to first calculate the second likelihood probability of each second data point in the second spectral feature data under each category. , calculate the overall second spectral feature data in each category The second joint likelihood probability under : Furthermore, according to Bayes' theorem, the second prior probability corresponding to each category is combined with the second joint likelihood probability corresponding to each category to calculate the second spectral posterior probability corresponding to each category. The specific calculation formula is: Step 302: Perform Bayesian fusion on the first spectral posterior probabilities corresponding to each category and the second spectral posterior probabilities corresponding to each category to calculate the fused posterior probabilities of each category.

[0079] In this embodiment, step 302 is used to fuse the first and second spectral posterior probabilities corresponding to each pigment category to obtain a fused posterior probability for each category. This step aims to maintain the advantages of independent modeling of the two types of spectral information while integrating their posterior outputs using a unified Bayesian decision fusion mechanism, thereby achieving a more robust and reliable classification of the pigment composition of the mural sample.

[0080] To achieve the above purpose, the fusion strategy adopted in this step is Bayesian multiplication fusion, that is, under the same category conditions, the first spectrum posterior probability and the second spectrum posterior probability are multiplied, and the product results of all categories are normalized to obtain the final fusion posterior probability distribution of each category. Specifically, suppose a candidate pigment category is , and its corresponding first spectrum posterior probability is , the corresponding second spectrum posterior probability is , then the calculation formula of the fusion posterior probability is as follows: The numerator in this formula represents the category The joint support strength when high posterior probabilities are obtained independently under the two spectral channels is obtained. The denominator is the normalized sum of the product results of all categories, ensuring that the fused posterior probability constitutes a legal probability distribution in all categories.

[0081] Step 303: Determine the pigment composition of the mural based on all fused posterior probabilities.

[0082] In this embodiment, step 303 determines the pigment composition of the mural based on the fused posterior probabilities and is the decision-making step in the overall recognition method for generating the final classification result. This step is implemented because, after calculating the posterior probabilities for the first and second spectral feature data and obtaining the fused posterior probability through Bayesian product fusion, the fused posterior probabilities for each category are analyzed to determine the pigment category most likely corresponding to the current sample being tested, thereby outputting a specific recognition result.

[0083] In a possible implementation, step 303 specifically includes steps 501 to 502: Step 501: Determine the maximum fused posterior probability from all fused posterior probabilities.

[0084] Step 502: Determine the pigment composition of the mural according to the pigment category corresponding to the maximum fused posterior probability.

[0085] Specifically, this step determines the maximum value of the fused posterior probability and determines the category with the maximum fused posterior probability as the target category to which the sample to be tested belongs. Suppose the candidate pigment categories in the fused posterior probability distribution are , and their corresponding fusion posterior probabilities are , then the actual judgment criteria implemented in this step are: , that is, select the one with the largest posterior probability among all candidate categories , and use it as the pigment composition category corresponding to the current sample.

[0086] This method has a clear probabilistic meaning: the category with the highest fused posterior probability represents the category with high support in both spectral feature channels, and therefore has higher credibility than other categories. This maximum a posteriori probability decision-making approach not only quantifies and automates classification decisions, but also ensures the interpretability and consistency of the decision logic. In practical applications, this method can effectively address complex situations with multi-category confusion or partial missing spectral features, prioritizing the most representative category from the fused information, thereby improving classification robustness.

[0087] Furthermore, because the fused posterior probability underlying this step is derived by normalizing the two spectral posterior probabilities using the Bayesian formula, it exhibits significant cross-modal consistency. This method can fully demonstrate its joint discriminative effect when different spectral channels have varying degrees of support, making the final output more comprehensive and informative.

[0088] In another possible implementation, step 103 may be further implemented through steps 601 to 603: Step 601: Determine all element types of the mural based on the first spectral characteristic data.

[0089] After determining the first spectral signature data, analysis is performed based on this data. This is done to fully utilize the first spectral signature data's ability to reflect elemental characteristics. By extracting these features and matching them with a database, the elemental composition of the sample can be accurately identified, providing a solid foundation for comprehensive, multi-layered pigment composition analysis.

[0090] In a possible implementation, step 601 specifically includes steps 701 to 702: Step 701: Match the first spectrum feature data with all first preset feature information in the preset standard element database one by one.

[0091] Step 702: When the first spectral characteristic data matches the first preset characteristic information, all element types of the mural are determined according to the element types corresponding to each matched first preset characteristic information.

[0092] Specifically, in step 701, a preset standard element database stores characteristic spectral data for a large number of known elements, including information such as characteristic wavelengths, peak intensity ranges, and peak widths. When matching all first spectral characteristic data against the first preset characteristic information for each element in the database, a specific matching algorithm is employed to compare the similarity between the position, intensity, and shape of the characteristic peaks and the standard information. If a particular first spectral characteristic data satisfies the specified matching criteria (e.g., position error within an allowable range, intensity within an expected range, etc.) with a particular first preset characteristic information in the database, the first spectral characteristic data is considered to correspond to the corresponding element in the database.

[0093] In practical implementation, matching accuracy and efficiency are crucial. To improve matching accuracy, the matching algorithm can be optimized, for example, using peak normalization and dynamic threshold adjustment to adapt to the characteristics of different peaks in the spectral data. Furthermore, by using a hierarchical screening matching process to prioritize characteristic peaks with a high matching probability, computational complexity can be reduced and matching efficiency improved.

[0094] By individually matching the first spectral signature data with a pre-set standard element database, all characteristic peaks extracted from the first actual spectral data can be associated with specific elemental species. This significantly improves the accuracy and comprehensiveness of sample element identification, especially in the analysis of multi-layered pigments or complex samples, ensuring that the elemental composition of each pigment layer is accurately identified.

[0095] Specifically, in step 701, a preliminary correlation is determined between some of the first spectral signature data and the first preset signature information in the preset standard element database through one-by-one matching. Next, in step 702, the element type corresponding to each first spectral signature data is extracted based on the matching results. During this process, the matching conditions are strictly adhered to the characteristics of the spectral signal. For example, the position deviation of the characteristic peak is within an allowable range (e.g., ±0.1nm), and the intensity is consistent with the standard value in the database or within a certain error range (e.g., ±10%). For each matching that meets the conditions, the corresponding element type is marked and recorded.

[0096] To improve efficiency, automated data processing tools can be used to generate a list of elemental species based on matching results, remove duplicates, and perform statistics to ensure that all elemental species are fully identified. Furthermore, for complex samples with potential signal overlap or multi-layer pigment interference, further screening can be performed by combining peak shape characteristics (such as peak width and symmetry) to eliminate possible false matches, thereby improving the accuracy of identification results.

[0097] By extracting the element types according to the matching results, the characteristic information hidden in the first actual spectrum data can be converted into clear chemical element types.

[0098] Step 602: Determine all molecular structure information of the mural based on all element types and the second spectral characteristic data.

[0099] After obtaining all elemental types and the second spectral signature data, analysis is performed on all of them. To filter out molecular composition information related to all elemental types, it is necessary to extract the second preset signature information corresponding to all elemental types from a preset molecular composition database. This is necessary because the signature information extracted from the second actual spectral data may contain mixed features of multiple molecular structures. By combining the identified elemental types, the database screening range can be effectively narrowed, thereby improving the efficiency and accuracy of molecular structure identification.

[0100] In a possible implementation, step 602 specifically includes steps 801 to 803: Step 801: Determine second preset characteristic information corresponding to all element types in a preset molecular component database according to all element types; each second preset characteristic information is associated with one element type.

[0101] Step 802: Match the second spectral characteristic data with all second preset characteristic information one by one.

[0102] Step 803: When the second spectral characteristic data matches all the second preset characteristic information, all the molecular structure information of the mural is determined according to the molecular structure information corresponding to each second preset characteristic information.

[0103] Specifically, in step 801, the preset molecular component database contains a wealth of molecular structure information. Each molecular structure is associated with a number of elemental species and stores its spectral characteristic information (i.e., second preset characteristic information), including characteristic peak position, intensity, and shape. During the screening process, based on all elemental species determined in step 702, candidate molecular structures associated with these elemental species are screened from the database, and the second preset characteristic information for each candidate is extracted.

[0104] The screening process is achieved by matching element types with molecular structure information in a molecular composition database. For example, if the identified elements include carbon (C), oxygen (O), and hydrogen (H), only the molecular structure information associated with C, O, and H is extracted from the database, while molecular structures associated with other elements (such as sulfur and chlorine) are ignored. For each screened molecular structure, the corresponding second preset feature information is extracted, which serves as the basis for the next step of spectral matching.

[0105] Through this screening and extraction process, the information in the pre-set molecular composition database is effectively trimmed, retaining only the information relevant to the actual element species. This not only reduces the computational effort for subsequent spectral matching but also significantly improves the accuracy and efficiency of molecular structure identification.

[0106] In step 802, by comparing the second spectral feature data one by one with the screened second preset feature information, it is possible to effectively verify whether the second spectral feature data is associated with a certain molecular component. During the matching process, the main features of the spectral peaks are compared, including parameters such as peak position, intensity and shape. First, by calculating the peak position difference between the second spectral feature data and the second preset feature information, ensure that the difference is within the allowable range (such as ±0.1nm). Secondly, compare the intensity of the corresponding peak to verify whether it meets the intensity range recorded in the preset feature information (such as the deviation is within 10%). Finally, check the shape parameters of the peak (such as peak width, symmetry) to ensure that the signal shape is consistent with the preset features in the database.

[0107] To further improve efficiency and accuracy during the matching process, the primary characteristic peaks (i.e., those with greater peak intensities) within the second spectral signature data can be prioritized, followed by matching of secondary characteristic peaks, gradually refining the matching results. If a particular second spectral signature data completely matches a second preset characteristic information in the database, the second spectral signature data is determined to correspond to the molecular structure associated with the preset characteristic information.

[0108] In step 802, the matching results of each second spectral feature data set against the second preset feature information are recorded one by one, including whether the match was successful and the corresponding molecular structure information. In step 803, these matching results are comprehensively analyzed to determine the overall matching status of the second actual spectral data. If all second spectral feature data sets can be matched with the second preset feature information, and these matches are consistent with the molecular structure information in the database, it indicates that the characteristic signals in the second actual spectral data can be fully explained by the screened molecular components.

[0109] During the analysis process, the reliability of the matching results can be further verified by statistically analyzing the degree of matching for each molecular structure. For example, when the information of a molecular structure is matched by multiple secondary spectral feature data, and its primary and secondary peaks are consistent with the database records, the molecular structure is considered highly reliable. In the case of incomplete matches, by examining the unmatched feature information, it is possible to determine whether there is signal noise or molecular structures not included in the database, providing a basis for subsequent database updates or supplementary analysis.

[0110] Through this comprehensive matching and verification process, all molecular structures that match the spectral signatures are compiled into a molecular structure list of the mural. This list fully reflects the chemical composition of the sample, encompassing the molecular components of each layer within the multi-layered pigment.

[0111] Step 603: Determine the pigment composition of the mural based on all element types and all molecular structure information.

[0112] To ultimately determine the mural's pigment composition during step 603, comprehensive information on all element types, all element contents, and all molecular structures is required. By organically combining these three dimensions of information, a comprehensive analysis of the pigment's chemical composition and its hierarchical distribution can be achieved, providing more accurate and reliable results, particularly for the analysis of multi-layered pigment samples. The goal is to integrate the multidimensional spectral information extracted from the sample into an accurate estimate of the pigment's composition, achieving a comprehensive analysis of the sample's composition.

[0113] Specifically, first, based on all the identified element types, candidate pigments in the pigment database that may match these elements are screened. This step performs preliminary filtering by comparing the chemical composition of the pigment with the extracted element types, eliminating pigment types that are obviously inconsistent. Next, the range of candidates is further narrowed down using the total element content. For each candidate pigment, the absolute value or ratio of the element content is compared with the component ratio of the standard pigment in the database to ensure that the screened pigments are consistent not only in element type, but also in content distribution with the actual sample. For example, if the lead content in the sample is significantly higher than other elements, it is more likely to be lead white rather than other lead-containing pigments.

[0114] After completing the screening based on element type and content, the final match is performed by combining all the determined molecular structure information. By comparing molecular structural features such as molecular vibration modes and the position of the main spectral peaks, the molecular composition of the candidate pigment is further verified to be consistent with the sample. During this process, the primary component of the candidate pigment is prioritized, while the rationality of the secondary components is verified to ensure the accuracy and completeness of the matching results.

[0115] For multi-layered pigment samples, this matching analysis is performed layer by layer. The possible pigment composition of each layer is determined based on the element types, element content, and molecular structure information. Comprehensive analysis of the matching results at different levels can reveal the historical evolution of the multi-layered pigment or the relationships between the layers.

[0116] Ultimately, by combining all element types, total element content, and complete molecular structure information, a report on the pigment composition of the mural is generated. This report includes the specific pigment name, chemical composition of each pigment, element content and ratio, and the distribution of pigment composition in each layer of the multi-layer sample.

[0117] Figure 2 This is one of the external structural schematic diagrams of the mural pigment detection device based on multi-spectral analysis provided by the present invention; Figure 3 This is the second schematic diagram of the external structure of the mural pigment detection device based on multi-spectral analysis provided by the present invention. Figure 2 and Figure 3 As shown, the mural pigment detection device based on multi-spectral analysis includes: Housing 0, switch button 1, display screen 2, grip 3, battery 4, data interface 5, trigger 6 and light outlet 7; The switch button 1 is arranged on the outer surface of the shell 0; the display screen 2 is embedded in the outer surface of the shell 0; the upper end of the handle 3 is fixedly connected to the shell 0; the battery 4 is detachably installed at the lower end of the handle 3; the data interface 5 is arranged on the outer surface of the shell 0; the trigger 6 is arranged at the front end of the handle 3; and the light outlet 7 is opened at the front end of the shell 0.

[0118] In this embodiment, a mural pigment detection device based on multispectral analysis is provided. Its structural design is designed to meet the needs of portable on-site operation and achieve efficient and stable spectroscopic detection performance. The device primarily comprises a housing 0, a switch button 1, a display 2, a grip 3, a battery 4, a data interface 5, a trigger 6, and a light exit 7. Its overall layout is compact and rational, making it easy for the user to hold in one hand and perform high-frequency detection.

[0119] To facilitate human-computer interaction, a power button 1 is located on the outer surface of the housing 0. The user can quickly activate and deactivate the device by long-pressing it, simplifying the operation process and preventing accidental touches. The display screen 2 is embedded in the outer surface of the housing 0. Its ergonomically optimized position allows the user to directly view the contents of the display screen 2 while holding the detection device, enabling rapid issuance of detection commands and intuitive presentation of results. The display screen 2 supports touch operation and, in conjunction with the internal control system, can perform operations such as spectrum display, function menu switching, and laser control, thereby enhancing the device's operability and intelligence.

[0120] The upper end of the handle 3 is fixedly connected to the housing 0, employing an integrated molded structure to ensure the device's overall mechanical strength and structural stability. The shape of the handle 3 conforms to the curve of the palm, enhancing user comfort during prolonged grip. The battery 4 is removably mounted at the lower end of the handle 3, allowing users to quickly replace the battery 4 without interrupting the testing process, significantly improving the device's battery life and field adaptability.

[0121] The data interface 5 is provided on the outer surface of the housing 0 and preferably adopts a Type-C interface structure, which provides bidirectional data transmission and charging functions, saving space while achieving multi-functional integration. This interface supports connection to a USB flash drive, computer, or other host system, facilitating the import and export of test data, improving information flow efficiency, and supporting external power supply to ensure continuous operation of the device.

[0122] Trigger 6, located at the front end of grip 3, is designed to fit the natural placement of the index finger, ensuring precise control of laser emission commands while holding the device, thereby triggering the detection process. This trigger 6 works in conjunction with the control circuit to control the emission of the Raman or LIBS light source based on the pressing state, balancing ease of use with safety.

[0123] Light exit aperture 7, located at the front end of housing 0 and coaxially aligned with the internal Raman probe 13 and LIBS optical path, projects a laser beam onto the surface of the mural sample, achieving spectral excitation at the target location. Light exit aperture 7 is centrally located and shielded to effectively prevent ambient light interference and control the laser divergence angle, thereby improving detection accuracy and stability.

[0124] Through the above-mentioned structural configuration and functional integration, the device not only has a good handheld operation experience and efficient spectral detection capabilities, but also has good portability.

[0125] Figure 4 This is one of the internal structure diagrams of the mural pigment detection device based on multi-spectral analysis provided by the present invention; Figure 5 This is the second schematic diagram of the internal structure of the mural pigment detection device based on multi-spectral analysis provided by the present invention; Figure 6This is the third schematic diagram of the internal structure of the mural pigment detection device based on multi-spectral analysis provided by the present invention; Figures 4 to 6 As shown, the mural pigment detection device based on multi-spectral analysis also includes: a first controller 8, a second controller 9, a first spectrometer 10, a continuous laser 11, a second spectrometer 12, a Raman probe 13, a Raman fiber connection end 14, a continuous laser fiber connection end 15, an N20 reduction motor 16, a pulse laser 17, a screw 18, a ball guide rail 19, a LIBS light output hole 20, a Raman light transceiver hole 21 and a LIBS light receiving hole 22.

[0126] The first controller 8, the second controller 9, the first spectrometer 10, the continuous laser 11, the second spectrometer 12, the Raman probe 13, the Raman fiber connection end 14, the continuous laser fiber connection end 15, the N20 reduction motor 16, the pulse laser 17, the screw 18, the ball guide 19, the LIBS light output hole 20, the Raman light receiving hole 21 and the LIBS light receiving hole 22 are all fixedly installed inside the housing 0; The first controller 8 is electrically connected to the display screen 2 and is in communication with the second controller 9; The second controller 9 is electrically connected to the continuous laser 11 and the pulse laser 17 respectively to control the light output of the continuous laser 11 and the pulse laser 17 respectively; The continuous laser 11 is connected to the Raman probe 13 via the continuous laser fiber connection end 15; The Raman probe 13 is set on the ball guide rail 19 and moves back and forth along the screw 18 under the drive of the N20 reduction motor 16; The Raman light transceiver hole 21 is provided at the front end of the Raman probe 13. The Raman probe 13 is used to transmit continuous laser light to the sample surface through the Raman light transceiver hole 21 and receive Raman light signals. The LIBS light exit hole 20 is provided at the front end of the pulse laser 17. The pulse laser 17 is used to emit LIBS light to the sample surface through the LIBS light exit hole 20 and receive the LIBS light signal after passing through the sample surface through the LIBS light receiving hole 22. The first spectrometer 10 and the second spectrometer 12 are connected to the Raman fiber connection end 14 and the LIBS light receiving hole 22 respectively through Z-type optical fibers. The Z-type optical fibers include a branching structure that splits the LIBS optical signal into two paths, one of which directly enters the second spectrometer 12, and the other merges with the Raman optical signal and enters the first spectrometer 10. Output ends of the first spectrometer 10 and the second spectrometer 12 are both electrically connected to the first controller 8 for transmitting spectral data to the display screen 2 for display.

[0127] In this embodiment, the mural pigment detection device based on multi-spectral analysis not only has a good human-computer interaction structure, but also integrates multiple sets of laser excitation, spectrum acquisition and control operation components inside its shell 0, which can simultaneously support Raman spectroscopy analysis and laser-induced breakdown spectroscopy (LIBS) analysis, thereby improving the component identification ability and detection accuracy of the detection object.

[0128] The device includes: a first controller 8, a second controller 9, a first spectrometer 10, a continuous laser 11, a second spectrometer 12, a Raman probe 13, a Raman fiber connection end 14, a continuous laser fiber connection end 15, an N20 reduction motor 16, a pulse laser 17, a screw 18, a ball guide rail 19, a LIBS light output hole 20, a Raman light transceiver hole 21 and a LIBS light receiving hole 22. The above components are all fixedly installed inside the shell 0, so that the entire device forms an integrated detection structure, which meets the dual requirements of portability and functional integration in the field of mural cultural relics protection.

[0129] To enable data acquisition, display, and logical control during the detection process, the device provides a communication connection between a first controller 8 and a second controller 9. First controller 8 is electrically connected to display screen 2 via a DSI signal interface, enabling real-time output and graphical presentation of spectral data. This allows users to view Raman or LIBS spectrum curves, adjust parameters, and input detection instructions.

[0130] The second controller 9 is responsible for accurately controlling the laser system. It is electrically connected to the continuous laser 11 and the pulsed laser 17, and sends light emission commands through the serial communication protocol to realize functions such as starting the laser, power regulation, and timing emission. Among them, the continuous laser 11 is used to excite the Raman signal, while the pulsed laser 17 is used to excite the plasma to generate the LIBS signal. The two correspond to different spectral analysis mechanisms. To achieve effective transmission of the laser, the light-emitting end of the continuous laser 11 is connected to the Raman probe 13 through the continuous laser fiber connection end 15. This fiber connection structure not only improves the laser transmission efficiency, but also enhances the flexibility of the system structure and the replaceability of the modules.

[0131] The Raman probe 13 is mounted on a ball-beam guide rail 19 and cooperates with a screw 18 to form a linear motion mechanism. One end of the screw 18 is linked to an N20 reduction motor 16 installed within the housing 0, which drives the Raman probe 13 forward and backward along the guide rail. This structural arrangement primarily addresses issues such as insufficient working distance of the Raman probe 13 in a compact structure and obstruction of the LIBS optical path. During Raman detection, the motor controls the probe to move forward to the working focus; during LIBS detection, the motor drives the probe backward to the storage position, ensuring that the two detection paths do not interfere with each other and enabling dynamic switching and coordinated operation of the detection process.

[0132] A Raman optical transceiver port 21, located at the front end of the Raman probe 13, is used to converge continuous laser light onto the surface of the mural sample. It also receives Raman signals scattered back from the sample surface and guides them to the fiber outlet via an internal reflection structure. From there, they are transmitted to the back-end spectrometer via the Raman fiber connector 14. Simultaneously, a LIBS light exit port 20, located at the corresponding position of the pulsed laser 17's light exit, precisely directs high-energy pulsed laser light onto the sample surface, stimulating the LIBS signal emitted by the plasma. After the signal is collected, it enters the fiber path through the LIBS light receiving port 22 located at the front of the device for further analysis.

[0133] To achieve shared detection and wavelength expansion of the two optical signals, the device utilizes a Z-type fiber connection structure. The Z-type fiber has a branching function, splitting the signal from the LIBS light receiving aperture 22 into two. One path is directly input into the second spectrometer 12 for high-sensitivity element identification; the other path is combined with the Raman signal transmitted by the Raman fiber in a coupler to form a single path, which is then fed into the first spectrometer 10, multiplexing the two spectroscopic signals within a single detection path. This multiplexing structure not only improves the device's structural integration but also expands the LIBS detectable wavelength range, enhancing the overall identification capability of metals, non-metals, and complex pigment materials.

[0134] In terms of data output, the output ends of the first spectrometer 10 and the second spectrometer 12 are both electrically connected to the first controller 8. The collected spectral information is uniformly processed and formatted by the first controller 8, and finally the spectrum is presented through the display screen 2 for real-time viewing by the user.

[0135] Further, refer to Figure 7 and Figure 8 , Figure 7 The mural pigment detection device provided by the present invention is used for the atomic emission spectrometry detection of realgar and orpiment pigments. Figure 8 The present invention provides a mural pigment detection device for detecting realgar and orpiment pigments using Raman spectroscopy. The mural pigment detection device is a Raman and LIBS combined device.

[0136] like Figure 7 The figure shows the atomic emission spectrum of the Raman and LIBS combined device used for the detection of realgar and orpiment pigments. Figure 7 It can be found that the atomic emission spectra of orpiment and realgar are the same. According to the standard spectrum, the characteristic element is As, which is mainly atomic lines. The stronger characteristic peaks of As are 228.85nm, 235.01nm, 236.98nm, 245.66nm, 278.03nm286.05nm. There are also atomic emission lines of trace elements Fe, Mg, Si, Ca, etc.

[0137] like Figure 8 The figure shows the Raman spectrum of the Raman and LIBS combined device of the present invention for the detection of realgar and orpiment pigments. In this example, the LIBS spectra of the two are basically the same, and it is difficult to distinguish them by LIBS detection. However, there are obvious differences in the molecular structures of realgar and orpiment, which causes them to show significantly different characteristic peaks in the Raman spectrum. As can be seen from the figure, the Raman bands of the two are very different. The Raman peaks of realgar are 180cm-1, 196cm-1, 212cm-1, 270cm-1, 345cm-1, 354cm-1, and 365cm-1, and the corresponding vibration modes are: 212cm-1 belongs to As-As-S bending vibration, 180cm-1 belongs to As-S-As bending vibration, 270cm-1 belongs to As-S bending vibration, 345cm-1, 354cm-1 and 365cm-1 belong to As -S stretching vibration; the Raman peaks of orpiment are 175cm-1, 198cm-1, 288cm-1, 306cm-1, 349cm-1, and 378cm-1, and the corresponding vibration modes are: 198cm-1 belongs to As-S-As bending vibration, 175cm-1 belongs to S-As-S bending vibration, 288cm-1 and 306cm-1 belong to antisymmetric As-S stretching vibration, 349cm-1 belongs to As-S stretching vibration, and 378cm-1 belongs to As-S-As stretching vibration.

[0138] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for detecting mural pigments based on multispectral analysis, characterized in that: include: Acquiring first actual spectrum data and second actual spectrum data for characterizing components of the mural; extracting first spectrum characteristic data from the first actual spectrum data, and extracting second spectrum characteristic data from the second actual spectrum data; The pigment composition of the mural is determined according to the first spectral characteristic data and the second spectral characteristic data.

2. The mural pigment detection method based on multispectral analysis according to claim 1, characterized in that: The extracting first spectral characteristic data from the first actual spectral data and extracting second spectral characteristic data from the second actual spectral data includes: Calculating a first-order derivative and a second-order derivative of each first data point in the first actual spectrum data; determining at least one first characteristic peak in the first actual spectrum data according to changes in the first-order derivative and the second-order derivative of each of the first data points; Obtaining the first spectral characteristic data according to all the first characteristic peaks; Calculating a first-order derivative and a second-order derivative of each second data point in the second actual spectrum data; determining at least one second characteristic peak in the second actual spectrum data according to changes in the first-order derivative and the second-order derivative of each second data point; The second spectrum characteristic data is obtained based on all the second characteristic peaks.

3. The mural pigment detection method based on multispectral analysis according to claim 1 is characterized in that: The determining the pigment composition of the mural according to the first spectral characteristic data and the second spectral characteristic data includes: According to the first spectral feature data and the second actual spectral data, respectively obtaining a first spectral posterior probability of the first spectral feature data in each category and a second spectral posterior probability of the second spectral feature data in each category; Performing Bayesian fusion on the first spectral posterior probability corresponding to each category and the second spectral posterior probability corresponding to each category to calculate the fused posterior probability of each category; The pigment composition of the mural is determined based on all the fused posterior probabilities.

4. The mural pigment detection method based on multispectral analysis according to claim 3 is characterized in that: The obtaining, based on the first spectral feature data and the second actual spectral data, first spectral posterior probabilities of the first spectral feature data in each category and second spectral posterior probabilities of the second spectral feature data in each category, respectively, includes: Acquire first sample spectral data and second sample spectral data; Calculating a first priori probability of the first sample spectral data in each category, and calculating a second priori probability of the second sample spectral data in each category; Calculating a first mean and a first variance of the first sample spectral data and the first spectral feature data respectively, and calculating a second mean and a second variance of the second sample spectral data and the second spectral feature data respectively; Calculating a first likelihood probability of each first data point in the first spectral feature data under each category according to the first mean and the first variance; Calculating a second likelihood probability of each second data point in the second spectral feature data under each category according to the second mean and the second variance; Calculating a first spectral posterior probability corresponding to each category based on the first prior probability corresponding to each category and the first likelihood probability corresponding to each category; The second spectral posterior probability corresponding to each category is calculated according to the second prior probability corresponding to each category and the second likelihood probability corresponding to each category.

5. The mural pigment detection method based on multispectral analysis according to claim 3 is characterized in that: Determining the pigment composition of the mural according to all the fused posterior probabilities includes: determining a maximum fused posterior probability from all of the fused posterior probabilities; The pigment composition of the mural is determined according to the pigment category corresponding to the maximum fused posterior probability.

6. The method for detecting mural pigments based on multispectral analysis according to claim 1, characterized in that: The determining the pigment composition of the mural according to the first spectral characteristic data and the second spectral characteristic data further includes: determining all element types of the mural according to the first spectral characteristic data; determining all molecular structure information of the mural according to all the element types and the second spectral characteristic data; The pigment composition of the mural is determined based on all the element types and all the molecular structure information.

7. The method for detecting mural pigments based on multispectral analysis according to claim 6, characterized in that: The determining of all element types of the mural according to the first spectral characteristic data includes: Matching the first spectral characteristic data with all first preset characteristic information in a preset standard element database one by one; When the first spectral characteristic data matches the first preset characteristic information, all element types of the mural are determined according to the element types corresponding to each matched first preset characteristic information.

8. The method for detecting mural pigments based on multispectral analysis according to claim 6, characterized in that: Determining all molecular structure information of the mural based on all the element types and the second spectral characteristic data includes: Determining, based on all the element types, second preset characteristic information corresponding to all the element types in a preset molecular component database; each second preset characteristic information is associated with one element type; Matching the second spectral characteristic data with all the second preset characteristic information one by one; When the second spectral characteristic data matches all of the second preset characteristic information, all of the molecular structure information of the mural is determined according to the molecular structure information corresponding to each piece of the second preset characteristic information.

9. The method for detecting mural pigments based on multispectral analysis according to claim 1, characterized in that: Before obtaining the first actual spectral data and the second actual spectral data for characterizing the components of the mural, the method further includes: When the pulsed laser emits light, the laser-induced breakdown spectrum data is collected by a first spectrometer, and when the continuous laser emits light, the Raman spectrum data is collected by a second spectrometer; Normalization processing, smoothing and denoising processing, and spectral baseline correction processing are performed on the laser-induced breakdown spectroscopy data and the Raman spectroscopy data, respectively, to obtain the first actual spectral data and the second actual spectral data, respectively.

10. A mural pigment detection device based on multi-spectral analysis, characterized in that: include: Housing, switch button, display screen, grip, battery, data interface, trigger and light outlet; The switch button is arranged on the outer surface of the housing; The display screen is embedded in the outer surface of the housing; the upper end of the grip is fixedly connected to the housing; the battery is detachably mounted on the lower end of the grip; the data interface is provided on the outer surface of the housing; the trigger is provided at the front end of the grip; the light exit hole is provided at the front end of the housing; The mural pigment detection device based on multispectral analysis also includes: a first controller, a second controller, a first spectrometer, a continuous laser, a second spectrometer, a Raman probe, a Raman fiber connection end, a continuous laser fiber connection end, an N20 reduction motor, a pulse laser, a lead screw, a ball guide rail, a LIBS light output hole, a Raman light receiving hole, and a LIBS light receiving hole; The first controller, the second controller, the first spectrometer, the continuous laser, the second spectrometer, the Raman probe, the Raman fiber connection end, the continuous laser fiber connection end, the N20 reduction motor, the pulse laser, the lead screw, the ball guide rail, the LIBS light output hole, the Raman light transceiver hole, and the LIBS light receiving hole are all fixedly installed inside the housing; The first controller is electrically connected to the display screen and is in communication with the second controller; The second controller is electrically connected to the continuous laser and the pulse laser respectively to control the light output of the continuous laser and the pulse laser respectively; The continuous laser is connected to the Raman probe via the continuous laser optical fiber connection end; The Raman probe is arranged on the ball guide rail and moves forward and backward along the screw rod under the drive of the N20 reduction motor; The Raman light transceiver hole is provided at the front end of the Raman probe, and the Raman probe is used to transmit continuous laser light to the sample surface through the Raman light transceiver hole and receive Raman light signals; The LIBS light exit hole is provided at the front end of the pulse laser, and the pulse laser is used to emit LIBS light to the sample surface through the LIBS light exit hole, and receive the LIBS light signal after passing through the sample surface through the LIBS light receiving hole; The first spectrometer and the second spectrometer are respectively connected to the Raman fiber connection end and the LIBS light receiving hole through a Z-type optical fiber; wherein the Z-type optical fiber includes a branching structure, which splits the LIBS optical signal into two paths, one of which directly enters the second spectrometer, and the other is combined with the Raman optical signal and then enters the first spectrometer; The output ends of the first spectrometer and the second spectrometer are both electrically connected to the first controller for transmitting spectral data to the display screen for display.

Images