Stone cultural relic disease identification method based on hyperspectral image

Abstract

The invention discloses a stone cultural relic disease identification method based on a hyperspectral image, and aims to solve the problem that stone cultural relic disease investigation depends on artificial experience and is time-consuming and labor-consuming. The method comprises the steps that a hyperspectral image of a stone cultural relic to be explored is collected and preprocessed; extracting spectral features; selecting a central wave band of three wave bands of blue light, green light and red light to construct a true color image; extracting texture features and color features of the true color image; fusing the spectral features, the texture features and the color features to obtain combined features; establishing a typical disease feature database; training a disease classification model by using a random forest, a support vector machine, a gradient elevator and a convolutional neural network; carrying out typical disease identification by using the intelligent identification model and drawing a disease cloud picture; and calculating the typical disease area proportion of the to-be-detected area to carry out disease evaluation. According to the invention, in-situ, lossless and rapid identification of stone cultural relic diseases is realized, manpower and material resources are saved, and an intelligent identification method is provided for cultural relic protection exploration work.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of rock and soil cultural relics protection, and specifically to a method for identifying and classifying stone cultural relics diseases based on hyperspectral. Background Art

[0002] Immovable stone heritage such as grottoes will suffer severe weathering due to long-term exposure to the natural environment, which is often accompanied by the appearance of peeling, salt crystallization, crusting, and biological colonization, thus causing the destruction of the structural integrity and value of stone cultural relics. Therefore, the study of the identification of stone cultural relics is the first important step in the targeted protection of stone cultural relics.

[0003] However, the current method for identifying the damage of rock cultural relics is for staff to take optical photos of the area to be tested and then use statistical software to delineate the affected area. This depends entirely on the professional level of the staff, the learning cost is too high and the identification is not accurate enough. Therefore, a fast, accurate and low-cost method for identifying the disasters of stone cultural relics is urgently needed. Summary of the invention

[0004] In view of the above problems, an embodiment of the present invention provides a hyperspectral-based stone cultural relic disease identification and classification method, which is used to solve the problems of inefficiency, inaccuracy and high cost in traditional methods of stone cultural relic disaster identification.

[0005] According to one aspect of an embodiment of the present invention, a method for identifying stone cultural relics diseases based on hyperspectral is provided. Figure 1 As shown, the method includes:

[0006] Step 1: Collect the hyperspectral image of the stone cultural relics to be surveyed and perform preprocessing;

[0007] Step 2: Extract the spectral features of the preprocessed hyperspectral image;

[0008] Step 3: Select the central bands of the three bands of blue, green and red light from the preprocessed hyperspectral image to construct a true color image;

[0009] Step 4: Extract texture features and color features of true color images;

[0010] Step 5: The spectral features, texture features and color features of the preprocessed hyperspectral image are fused using a weighted combination method to obtain a combined feature;

[0011] Step 6: Label the samples of the characteristics of four different disease types, namely, spalling, salt crystallization, biological colonization, and crusting, and establish a typical disease characteristic database;

[0012] Step 7: Using the combined feature vector as input, train the disease classification model using random forest, support vector machine, gradient boosting machine, and convolutional neural network;

[0013] Step 8: Use the intelligent recognition model to identify typical diseases and draw a disease cloud map;

[0014] Step 9: Conduct disease assessment by calculating the percentage of typical disease areas in the tested area.

[0015] In an optional manner, a hyperspectral imager is used to collect hyperspectral data of the area to be detected of the stone cultural relics in step 1. The hyperspectral imager has the advantages of strong continuity of spectral information, chemical composition recognition ability, high spatial resolution, and non-contact imaging.

[0016] In an optional manner, the preprocessing work in step 1 includes processing the hyperspectral image using a Savitzky-Golay smoothing filter and normalization processing. The Savitzky-Golay smoothing filter has the advantages of effectively suppressing noise, retaining image features, being able to adapt to image processing requirements, and having a simple and efficient algorithm.

[0017] In an optional manner, in step 2, the positions and numbers of absorption peaks on the spectrum curve within the wavelength range of 400-2500 nm are analyzed as spectral features of the disease, and the types of diseases are distinguished thereby.

[0018] In an optional manner, the absorption characteristics of the spectral curve are that exfoliation and salt crystallization have multiple obvious absorption peaks near 700-1000nm and 1400nm; the spectral curve of biological colonization and crusting has an absorption peak near 550nm (nitrogen absorption band) and an obvious rising stage at 700-750nm (light scattering inside leaf cells); the spectral reflectivity of salt crystallization will be higher than other disaster types.

[0019] In an optional manner, in step 3 or step 4, three bands of blue light, green light and red light centers (475, 520 and 650 nm) are selected from the hyperspectral image of the stone cultural relics in the test area to construct a true color image to extract texture features and color features.

[0020] In an optional manner, the light with wavelengths of 475, 520 and 650 nm in step 3 or step 4 has high resolution and contrast in the spectral image, so that the color details in the image can be more accurately reflected when extracting features.

[0021] In an optional manner, the color feature in step 4 is obtained by converting the RGB color model into the HSV color model. The principle of color space conversion is:

[0022] R′=R / 255

[0023] G′=G / 255

[0024] B′=B / 255

[0025] C max =max(R′, G′, B′)

[0026] C min =min(R′, G′, B′)

[0027] Δ=C max -C min

[0028] The calculation formula of hue H is:

[0029]

[0030] The calculation formula of saturation S is:

[0031]

[0032] The calculation formula of brightness V is:

[0033] V=C max

[0034] In an optional manner, the texture extraction based on the gray co-occurrence matrix (GLCM) in step 4 is obtained by calculating the joint conditional probability density between different pixel intensities in the image. It provides four texture features: energy (Ene), contrast (Con), homogeneity (Hom) and entropy (Ent).

[0035] Energy: reflects the grayscale concentration of the image and the thickness of the texture. The thicker the texture, the greater the energy.

[0036]

[0037] Where p(i,j) is the element value in the constructed gray level co-occurrence matrix, and i and j represent the gray levels as indices.

[0038] Contrast: It reflects the clarity of the image and also the depth of the texture grooves.

[0039]

[0040] Where (i, j) represents the difference between the grayscale values ​​of two pixels.

[0041] Uniformity: reflects the texture similarity of the image. The higher the value, the more similar the local areas are and the smaller the grayscale difference is.

[0042]

[0043] Entropy: Indicates the randomness of the image content, but also reflects the amount of information and complexity of the image. If the texture is complex, the entropy value is greater.

[0044]

[0045] In an optional manner, the weighted combination method is used for fusion in step 5. Assume that the spectral feature of the hyperspectral image is X spec , the color feature is X col , the texture feature is X text , then the combined feature a can be expressed as follows:

[0046] a=[α·X spec ,β·X col ,γ·X text ]

[0047] Among them, α+β+γ=1.

[0048] In an optional manner, the disease database established in step 6 has the following specific structure, content and technical advantages:

[0049] Database structure: It adopts a hierarchical structure, which is divided into a main directory layer for disease types, a secondary directory layer for individual cultural relics, and a sub-directory layer for specific disease details of different parts of each cultural relic. This hierarchical structure facilitates the classification management and rapid retrieval of disease data of different stone cultural relics and different parts.

[0050] Database content: covers the spectral feature data, color feature data, texture feature data of various typical diseases of stone cultural relics, as well as detailed descriptions of the corresponding diseases (including the causes of the diseases, characteristics of the development stages, etc.), and also records statistical information such as the frequency of occurrence of each disease in different cultural relic cases. These rich data contents provide comprehensive data support for subsequent accurate disease classification, analysis and other operations.

[0051] Technical advantages: By integrating multi-dimensional disease feature data, compared with traditional single-feature databases, it can more accurately reflect the actual situation of the disease, improve the accuracy and reliability of subsequent operations such as disease identification, classification, and disease development trend prediction based on the database, and thus better serve the protection of stone cultural relics.

[0052] In an optional manner, in step 7, multiple algorithms are used to train the disease classification model. The specific algorithm application steps and combination methods are as follows:

[0053] Taking the combined feature vector as input, the classifiers were trained using random forest, support vector machine, gradient boosting machine, and convolutional neural network algorithms respectively;

[0054] Select samples that have not participated in modeling for classification testing, and input the test samples into the trained classifier for classification prediction.

[0055] Model performance evaluation: Use global accuracy (GA) and kappa coefficient to evaluate the performance of the classification model. GA and kappa coefficients are calculated through confusion matrix. The calculation formula is as follows:

[0056] Global Accuracy (GA):

[0057]

[0058] Where n is the total number of columns in the confusion matrix (total number of categories); x ii is the number of samples in the i-th row and i-th column of the confusion matrix (the number of correctly classified samples); N is the total number of samples used for accuracy evaluation.

[0059] Kappa coefficient:

[0060]

[0061] where x i+ and x +i are the total number of samples in the i-th row and i-th column respectively.

[0062] Through the above steps, four algorithms are used to build disease classification models respectively, and the model performance is evaluated to select the best degradation identification model.

[0063] In an optional manner, in step 8, it is necessary to perform degradation identification on the hyperspectral image based on the trained model to determine the area range of different degradation types (peeling, salt crystallization, biological colonization and crusting) in the image, that is, traverse each pixel point according to the size of the original image to classify and assign values, assign different types of diseases to different colors to obtain a disease cloud map.

[0064] In an optional manner, in step 9, the ratio of pixels of different colors in the cloud map is calculated and multiplied by the total area S of the region to obtain the typical disease area of ​​the stone cultural relics. The calculation formula is:

[0065] S i =α·S

[0066] In the formula, α is the area proportion of a certain disease, S is the area of ​​the region, and S i The area covered by a certain disease.

[0067] Compared with the prior art, the advantages and beneficial effects of the present invention are as follows: by collecting hyperspectral images of the area to be tested, extracting spectral features, texture features and color features, fusing the spectral features, texture features and color features of the hyperspectral images into a combined feature, thereby establishing a disease feature database, training a disease classification model and performing performance evaluation, and then performing disease classification or identification, a large amount of manpower and material resources used for disease exploration is saved, and it is no longer dependent on the experience of the staff, and disease identification statistics in microscopic areas can be more accurately achieved on a smaller scale.

[0068] The above description is only an overview of the technical solution of the embodiment of the present invention. In order to more clearly understand the technical means of the embodiment of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the embodiment of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are specifically listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] The accompanying drawings are only used to illustrate the embodiments and are not to be considered as limiting the present invention. In addition, the same reference symbols are used to represent the same components throughout the accompanying drawings. In the accompanying drawings:

[0071] Figure 1 A schematic diagram of a flow chart of an embodiment of the present invention is shown.

[0072] Figure 2 The specific flow chart of establishing the stone cultural relic disease identification model in the embodiment of the present invention is shown.

[0073] Figure 3 The figure shows the recognition result of the disease recognition model in the embodiment of the present invention.

[0074] Figure 4 The statistics of the area of ​​stone carving damage identified by the algorithm identification model in the embodiment of the present invention are shown. DETAILED DESCRIPTION

[0075] Exemplary embodiments of the present invention will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be limited to the embodiments set forth herein.

[0076] A method for identifying stone cultural relics diseases based on hyperspectral images, as shown in the attached Figure 1 and attached Figure 2 As shown, the following steps are included:

[0077] Step 1: Collect the hyperspectral image of the stone cultural relics to be surveyed and perform preprocessing;

[0078] In this step, a hyperspectral imager with a wavelength range of 400-2500nm is used to collect hyperspectral data of the area to be inspected for stone cultural relics. The spectral resolution is 5nm, and the hyperspectral image obtained by scanning has a total of 420 bands.

[0079] The preprocessing work in this step includes processing the hyperspectral image using a Savitzky-Golay smoothing filter and normalization.

[0080] Step 2: Extract the spectral features of the preprocessed hyperspectral image;

[0081] In this step, the positions and numbers of absorption peaks on the spectrum curve within the wavelength range of 400-2500nm are analyzed as the spectrum characteristics of the disease, and the types of diseases are distinguished accordingly.

[0082] The absorption characteristics of the spectral curve are that exfoliation and salt crystallization have multiple obvious absorption peaks near 700-1000nm and 1400nm; the spectral curve of biological colonization and crusting has an absorption peak near 550nm (nitrogen absorption band) and an obvious rising stage at 700-750nm (light scattering inside leaf cells); the spectral reflectivity of salt crystallization will be higher than other disaster types.

[0083] Step 3: Select the central bands of the three bands of blue, green and red light from the preprocessed hyperspectral image to construct a true color image;

[0084] In this step, the three bands of blue, green and red light centers (475, 520 and 650 nm) of the hyperspectral image of the stone cultural relics in the test area are selected to construct a true color image to extract texture features and color features.

[0085] Step 4: Extract texture features and color features of true color images;

[0086] In this step, the color features are obtained by converting the RGB color model to the HSV color model. The principle of color space conversion is:

[0087] R′=R / 255

[0088] G′=G / 255

[0089] B′=B / 255

[0090] C max =max(R′, G′, B′)

[0091] C min =min(R′, G′, B′)

[0092] Δ=C max -C min

[0093] The calculation formula of hue H is:

[0094]

[0095] The calculation formula of saturation S is:

[0096]

[0097] The calculation formula of brightness V is:

[0098] V=C max

[0099] In this step, texture extraction based on the grey co-occurrence matrix (GLCM) is obtained by calculating the joint conditional probability density between different pixel intensities in the image. It provides four texture features: energy (Ene), contrast (Con), homogeneity (Hom) and entropy (Ent).

[0100] Energy: reflects the grayscale concentration of the image and the thickness of the texture. The thicker the texture, the greater the energy.

[0101]

[0102] Where p(i,j) is the element value in the constructed gray level co-occurrence matrix, and i and j represent the gray levels as indices.

[0103] Contrast: It reflects the clarity of the image and also the depth of the texture grooves.

[0104]

[0105] Where (i, j) represents the difference between the grayscale values ​​of two pixels.

[0106] Uniformity: reflects the texture similarity of the image. The higher the value, the more similar the local areas are and the smaller the grayscale difference is.

[0107]

[0108] Entropy: Indicates the randomness of the image content, but also reflects the amount of information and complexity of the image. If the texture is complex, the entropy value is greater.

[0109]

[0110] Step 5: The spectral features, texture features and color features of the preprocessed hyperspectral image are fused using a weighted combination method to obtain a combined feature;

[0111] In this step, the weighted combination method is used for fusion. Assume that the spectral feature of the hyperspectral image is X spec , the color feature is X col, the texture feature is X text , then the combined feature a can be expressed as follows:

[0112] a=[α·X spec ,β·X col ,γ·X text ]

[0113] Among them, α+β+γ=1.

[0114] Step 6: Label the samples of the characteristics of four different disease types, namely, spalling, salt crystallization, biological colonization, and crusting, and establish a typical disease characteristic database;

[0115] Step 7: Using the combined feature vector as input, train the disease classification model using random forest, support vector machine, gradient boosting machine, and convolutional neural network;

[0116] In this step, multiple algorithms are used to train the disease classification model. The specific algorithm application steps and combinations are as follows:

[0117] Taking the combined feature vector as input, the classifiers were trained using random forest, support vector machine, gradient boosting machine, and convolutional neural network algorithms respectively;

[0118] Select samples that have not participated in modeling for classification testing, and input the test samples into the trained classifier for classification prediction.

[0119] Model performance evaluation: Use global accuracy (GA) and kappa coefficient to evaluate the performance of the classification model. GA and kappa coefficients are calculated through confusion matrix. The calculation formula is as follows:

[0120] Global Accuracy (GA):

[0121]

[0122] Where n is the total number of columns in the confusion matrix (total number of categories); x ii is the number of samples in the i-th row and i-th column of the confusion matrix (the number of correctly classified samples); N is the total number of samples used for accuracy evaluation.

[0123] Kappa coefficient:

[0124]

[0125] where x i+ and x +i are the total number of samples in the i-th row and i-th column respectively.

[0126] Through the above steps, four algorithms are used to build disease classification models respectively, and the model performance is evaluated to select the best degradation identification model.

[0127] Step 8: Use the intelligent recognition model to identify typical diseases and draw a disease cloud map;

[0128] In this step, it is necessary to identify the degradation of the hyperspectral image based on the trained model and determine the regional range of different degradation types (peeling, salt crystallization, biological colonization and crusting) in the image. That is, traverse each pixel point according to the size of the original image to classify and assign values, and assign different types of diseases to different colors to obtain a disease cloud map.

[0129] Step 9: Conduct disease assessment by calculating the percentage of typical disease areas in the tested area.

[0130] In this step, the ratio of pixels of different colors in the cloud map is calculated and multiplied by the total area S of the region to obtain the typical disease area of ​​stone cultural relics. The calculation formula is:

[0131] S i =α·S

[0132] In the formula, α is the area proportion of a certain disease, S is the area of ​​the region, and S i The area covered by a certain disease.

[0133] The following is an example of disease identification:

[0134] The Sakyamuni stone carvings of Dazu Rock Carvings are prone to various diseases due to long-term exposure to the natural environment, so timely disease identification is required. In order to reduce the error and noise in the spectral data acquisition process, 10 spectral images are collected in each area under the same setting conditions. The final spectral data can be obtained by arithmetic averaging.

[0135] A portable hyperspectral imager with a wavelength range of 400-2500nm was used to collect hyperspectral data of the area to be detected. The spectral resolution was 5nm, and the hyperspectral image obtained by scanning had a total of 420 bands. The hyperspectral image was processed and normalized using a Savitzky-Golay smoothing filter.

[0136] Then the spectral characteristics of the hyperspectral image are analyzed. Then the hyperspectral image is imported into the Analysis software, and 475, 520, and 650 nm are selected as the centers of the red, green, and blue light bands to derive the RGB image. The RGB image is then subjected to a series of calculations to obtain the HSV image in order to extract the color features of the hyperspectral image. The texture is described by calculating the joint conditional probability density between different pixel intensities in the image, thereby reflecting the correlation between the intensity values ​​of any two points and the statistical characteristics of the texture features. When extracting texture features, the size and direction of the moving window have little effect on the estimation. Therefore, a 5×5 window is selected in each cycle, and four texture features in the 45° direction are extracted for subsequent correlation analysis.

[0137] The weighted combination method is used to fuse the spectral features, texture features and color features of the hyperspectral image into a combined feature.

[0138] According to the observed disease types, the disease regions (ROIs) were delineated. This operation could be performed using ENVI5.6 software to establish a disease database.

[0139] The combined features were used as input vectors, and the disease classification model was trained using random forest, support vector machine, gradient boosting machine and convolutional neural network. The samples were divided into training set and test set in a ratio of 7:3.

[0140] The performance of the four disease recognition models is different for different disease types. By evaluating the classification model performance by global accuracy (GA) and kappa coefficient, the four disease recognition models showed high accuracy in the recognition performance of crusting and biological colonization, all above 0.90, so they can all recognize typical diseases. Among them, the gradient boosting model has the best comprehensive recognition performance (GA: 0.97; Kappa: 0.97).

[0141] The test samples were classified and identified for diseases. The identification results are shown in the attached Figure 3 ; By calculating the proportion of each type of disease in the tested area (area ratio), the results are shown in Figure 4 , so that experts can promptly propose specific restoration and maintenance plans to ensure the integrity and value of stone cultural relics.

[0142] In the description provided herein, a large number of specific details are described. However, it is understood that embodiments of the present invention can be practiced without these specific details. Similarly, in order to simplify the present invention and help understand one or more of the various inventive aspects, in the above description of exemplary embodiments of the present invention, the various features of the embodiments of the present invention are sometimes grouped together into a single embodiment, figure, or description thereof. Wherein, the claims that follow the specific embodiment are hereby expressly incorporated into the specific embodiment, wherein each claim itself is a separate embodiment of the present invention.

[0143] Those skilled in the art will appreciate that the modules in the devices in the embodiments may be adaptively changed and arranged in one or more devices different from the embodiments. The modules or units or components in the embodiments may be combined into one module or unit or component, and further may be divided into a plurality of submodules or subunits or subcomponents. Except that at least some of such features and / or processes or units are mutually exclusive.

[0144] It should be noted that the above embodiments illustrate the present invention rather than limit it, and that those skilled in the art may devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference symbol between brackets shall not be construed as a limitation on the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "one" or "an" preceding an element does not exclude the presence of a plurality of such elements. The present invention may be implemented by means of hardware comprising a number of different elements and by means of a suitably programmed computer. In a unit claim enumerating a number of devices, several of these devices may be embodied by the same hardware item. The use of the words first, second, and third, etc. does not indicate any order. These words may be interpreted as names. The steps in the above embodiments, unless otherwise specified, should not be understood as limitations on the order of execution.

Claims

1. A method for identifying stone cultural relics diseases based on hyperspectral images, characterized in that: The method comprises: Step 1: Collect the hyperspectral image of the stone cultural relics to be surveyed and perform preprocessing; Step 2: Extract the spectral features of the preprocessed hyperspectral image; Step 3: Select the central bands of the three bands of blue, green and red light from the preprocessed hyperspectral image to construct a true color image; Step 4: Extract texture features and color features of true color images; Step 5: The spectral features, texture features and color features of the preprocessed hyperspectral image are fused using a weighted combination method to obtain a combined feature; Step 6: Label the samples of the characteristics of four different disease types, namely, spalling, salt crystallization, biological colonization, and crusting, and establish a typical disease characteristic database; Step 7: Using the combined feature vector as input, train the disease classification model using random forest, support vector machine, gradient boosting machine, and convolutional neural network; Step 8: Use the intelligent recognition model to identify typical diseases and draw a disease cloud map; Step 9: Conduct disease assessment by calculating the percentage of typical disease areas in the tested area.

2. The method according to claim 1, characterized in that In the step 1, a hyperspectral imager is used to collect hyperspectral data of the area to be inspected of the stone cultural relics.

3. The method according to claim 1, characterized in that In step 3 or step 4, three bands of blue light, green light and red light center (475, 520 and 650 nm) are selected from the hyperspectral image of the stone cultural relics in the test area to construct a true color image to extract texture features and color features.

4. The method according to claim 1, characterized in that In step 3 or step 4, the light with wavelengths of 475, 520 and 650 nm has high resolution and contrast in the spectral image, so that the color details in the image can be more accurately reflected when extracting features.

5. The method according to claim 1, characterized in that In the step 4, the image is first converted from the RGB color space to the HSV color space, and color features are extracted under the HSV color model, specifically, the proportion of the number of pixels in a specific hue interval in the H channel, the proportion of the number of pixels in a specific saturation interval in the S channel, and the proportion of the number of pixels in a specific brightness interval in the V channel are counted; at the same time, texture features are obtained through the grayscale co-occurrence moment (GLCM), and parameters such as the contrast, correlation, energy and entropy of the GLCM are calculated; then the above color feature parameters and texture feature parameters are combined into a feature vector, which is input into a pre-trained disease recognition classifier, and the presence or absence of a disease and the type of disease are determined based on the output result of the classifier, thereby realizing the application and combination of color features and texture features in disease recognition.

6. The method according to claim 1, characterized in that In step 5, weighted combination method is used for fusion. Assume that the spectral feature of the hyperspectral image is X spec , the color feature is X col , the texture feature is X text , then the combined feature a can be expressed as follows: a=[a·X spec ,b·X col ,γ·X text ] Among them, α+β+γ=1.

7. The method according to claim 1, characterized in that The disease database established in step 6 has the following specific structure, content and technical advantages: Database structure: It adopts a hierarchical structure, which is divided into a main directory layer for disease types, a secondary directory layer for individual cultural relics, and a sub-directory layer for specific disease details of different parts of each cultural relic. This hierarchical structure facilitates the classification management and rapid retrieval of disease data of different stone cultural relics and different parts. Database content: covers the spectral feature data, color feature data, texture feature data of various typical diseases of stone cultural relics, as well as detailed descriptions of the corresponding diseases (including the causes of the diseases, characteristics of the development stages, etc.), and also records statistical information such as the frequency of occurrence of each disease in different cultural relic cases. These rich data contents provide comprehensive data support for subsequent accurate disease classification, analysis and other operations. Technical advantages: By integrating multi-dimensional disease feature data, compared with traditional single-feature databases, it can more accurately reflect the actual situation of the disease, improve the accuracy and reliability of subsequent operations such as disease identification, classification, and disease development trend prediction based on the database, and thus better serve the protection of stone cultural relics.

8. The method according to claim 1, characterized in that: In step 7, multiple algorithms are used to train the disease classification model. The specific algorithm application steps and combination methods are as follows: Taking the combined feature vector as input, the classifiers were trained using random forest, support vector machine, gradient boosting machine, and convolutional neural network algorithms respectively; Select samples that have not participated in modeling for classification testing, and input the test samples into the trained classifier for classification prediction. Model performance evaluation: Use global accuracy (GA) and kappa coefficient to evaluate the performance of the classification model. GA and kappa coefficients are calculated through confusion matrix. The calculation formula is as follows: Global Accuracy (GA): Where n is the total number of columns in the confusion matrix (total number of categories); x ii is the number of samples in the ith row and ith column of the confusion matrix (the number of correctly classified samples); N is the total number of samples used for accuracy evaluation. Kappa coefficient: where x i+ and x +i are the total number of samples in the i-th row and i-th column respectively. Through the above steps, four algorithms are used to build disease classification models respectively, and the model performance is evaluated to select the best degradation identification model.

9. The method according to claim 1, characterized in that: In step 8, it is necessary to identify the degradation of the hyperspectral image based on the trained model, determine the area range of different degradation types (peeling, salt crystallization, biological colonization and crusting) in the image, that is, traverse each pixel point according to the size of the original image to classify and assign values, and assign different types of diseases to different colors to obtain a disease cloud map.

10. The method according to claim 1, characterized in that In step 9, the ratio of pixels of different colors in the cloud map is calculated and multiplied by the total area S of the region to obtain the typical disease area of ​​the stone cultural relics. The calculation formula is: S i =α·S In the formula, α is the area proportion of a certain disease, S is the area of ​​the region, and S i The area covered by a certain disease.

Images