Digital restoration and enhancement method and system for ancient painting books

Abstract

This invention discloses a method and system for digital restoration and enhancement of ancient Chinese paintings, belonging to the field of image processing. The invention adjusts the initial separation influence weights based on acquired spectral reflectance data to obtain optimized separation influence weights. Then, features are extracted from each candidate band and combined with the optimized separation influence weights to obtain a discrimination ratio. It is then determined whether this ratio is higher than a separation threshold. If not, candidate bands are adjusted to obtain optimized candidate bands, and their discrimination ratios are again checked against the threshold. Otherwise, the current candidate bands are output. The center position of each band is then determined for ink and pigment layer separation. The separation effect is assessed to determine whether the optimized separation influence weights need further adjustment. Finally, after separating the ink and pigment layers based on the band center position, digital restoration and enhancement are performed. This method enhances the stability and accuracy of ink and pigment layer separation, solving the problem of decreased separation accuracy in existing technologies.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to a method and system for the digital restoration and enhancement of ancient Chinese paintings and books. Background Technology

[0002] Digital restoration and enhancement of traditional Chinese paintings and ancient books is a cutting-edge technological direction in the field of cultural heritage protection and dissemination. It combines traditional archaeology, digital image processing, computer vision and artificial intelligence and other multidisciplinary technologies to maximize the protection, restoration and utilization of these precious historical and cultural resources.

[0003] Currently, the digital restoration of ancient books and traditional Chinese paintings presents two trends: on the one hand, the integration of traditional craftsmanship and digital technology. Traditional restoration remains fundamental, such as physical processes like paper repair and color restoration, but digital technology has become an important means to assist and improve efficiency. On the other hand, artificial intelligence is being introduced into the digital restoration process. AI (Artificial Intelligence) models (such as deep learning and generative models) are being used for tasks such as damage identification, missing parts replacement, and fading restoration, significantly improving the effectiveness and efficiency of automated restoration.

[0004] For example, Chinese invention patent CN120765513B discloses a method and system for batch digital restoration of ancient books, which includes: scanning the ancient book to be restored and generating multiple original images of the ancient book based on all its pages; performing image preprocessing on the multiple original images to generate multiple digital images of the ancient book, including noise detection, normalization, and format conversion; generating spatial feature maps and content feature maps based on the multiple digital images of the ancient book; determining a joint restoration decision based on the spatial feature maps and content feature maps, and batch restoring the multiple digital images of the ancient book based on the joint restoration decision. By connecting edges, the discrete regions to be restored are organized into an organic whole.

[0005] For example, Chinese invention patent CN119399073B discloses a method, apparatus, device, and medium for digital restoration of ancient Chinese book images, including: acquiring an image of an ancient book with missing text, and the original text data corresponding to the image; using a preset Chinese character detection model to obtain the coordinates of the positioning boxes of all characters on the ancient book image, checking the missing positions on the original image based on the positioning box coordinates and the original text data, and obtaining the positions of the characters that need to be repaired; obtaining the writing style in the ancient book image, generating text images with the corresponding writing style according to the characters to be repaired; performing pixel-level segmentation processing on the generated text, and pasting the text images onto the corresponding positions on the ancient book image according to the obtained positions of the repaired characters.

[0006] Current technologies for the digital restoration of traditional Chinese paintings and ancient books typically include the following key technical modules: a high-precision digital acquisition module, which requires high-resolution scanning or photography of paper books and paintings to obtain digital materials of the original works; an image enhancement and preprocessing module, such as noise reduction, contrast adjustment, color standardization, and pose and geometric correction; a damage detection and classification module, which uses computer vision and machine learning to detect typical damage such as cracks, insect infestation, stains, and mold, and AI models automatically label damaged areas to provide location for subsequent restoration; an AI-assisted restoration module, which uses neural network models to automatically perform "virtual restoration" of missing areas, image completion to automatically infer the image content of missing parts, similar to color filling and reconstruction, color restoration to repair faded color areas, and style constraint models to ensure that the restored brushstrokes maintain stylistic consistency with the original work. For example, recent research has proposed AI algorithms such as ACP-LaMa for ancient Chinese paintings, which can improve the artistry and quality of restoration. In addition, for documents that are difficult to open, such as scrolls and bamboo slips, it may also be necessary to adopt virtual unfolding technology, OCR (Optical Character Recognition) text recognition and semantic annotation, as well as knowledge graphs and semantic analysis technologies.

[0007] The above-mentioned technology has at least the following technical problems:

[0008] In existing technologies, the pigments in traditional Chinese paintings and ancient books are mostly natural mineral or plant pigments (cinnabar, azurite, malachite, ochre, indigo, etc.), rather than modern chemical pigments. These pigments have extremely different reflectance spectra and undergo chemical changes over time. The non-linear distribution of the spectral curves of different pigments leads to a decrease in color resolution. Some pigments (such as cinnabar) will "darken" or "turn black," making digital restoration difficult and reducing the precision of separating the ink layer and pigment layer during the restoration process. Summary of the Invention

[0009] To address the technical problem of decreased separation accuracy between the ink layer and pigment layer in existing technologies, this invention provides a method and system for the digital restoration and enhancement of ancient Chinese paintings. The technical solution is as follows:

[0010] On the one hand, a method for digital restoration and enhancement of ancient Chinese paintings is provided. This method includes: acquiring a spectral sample set of ink layer and a spectral sample set of pigment layer, along with their corresponding spectral wavelengths, and calculating the separation difference to obtain the separation difference degree; simultaneously acquiring the spectral reflectance data of the pigment layer spectral sample set, and obtaining the initial separation influence weight of the pigment layer from a preset database; constructing a difference-wavelength variation curve based on the spectral wavelength and the corresponding separation difference degree, and selecting candidate wavelengths accordingly, where candidate wavelengths represent the spectral wavelength range covered by the separation difference degree higher than the difference threshold in the difference-wavelength variation curve; adjusting the initial separation influence weight based on the pigment aging degree reflected by the spectral reflectance data to obtain an optimized separation influence weight; and extracting features from each candidate wavelength to obtain an ink layer feature vector. The system calculates the discrimination ratio by combining the feature vector of the pigment layer with the weights for optimizing the separation effect. The discrimination ratio measures the influence of the current candidate band on the separation of the ink and pigment layers. It then determines whether the discrimination ratio is higher than the preset separation threshold. If it is, the current candidate band is output; otherwise, the candidate band is adjusted until the discrimination ratio is higher than the separation threshold, and the candidate band is output. The center position of the band is determined by using the light source constraint of the output candidate band and the target sampling device, and the ink and pigment layers are separated. The separation effect is judged based on the separation effect evaluation index and the reference effect discrimination data, and the decision is made on whether to adjust the weights for optimizing the separation effect based on the judgment result. The ink and pigment layers of the ancient Chinese painting are separated according to the center position of the band, and then digital restoration and enhancement are performed.

[0011] On the other hand, a digital restoration and enhancement system for ancient Chinese paintings is provided, comprising: a digital data acquisition module for acquiring ink layer spectral sample sets and pigment layer spectral sample sets and their corresponding spectral wavelengths, and performing difference calculations to obtain the separation difference degree; simultaneously acquiring spectral reflectance data of the pigment layer spectral sample set, and obtaining the initial separation influence weight of the pigment layer from a preset database; a digital data processing module for constructing a difference-wavelength variation curve based on the spectral wavelength and the corresponding separation difference degree, and thereby selecting candidate wavelengths, where candidate wavelengths represent the spectral wavelength range covered by the difference-wavelength variation curve where the separation difference degree is higher than the difference threshold; a pigment layer weight adjustment module for adjusting the initial separation influence weight according to the pigment aging degree reflected by the spectral reflectance data to obtain an optimized separation influence weight; and a candidate wavelength separation effect evaluation module for extracting features from each candidate wavelength. The system obtains the feature vectors of the ink layer and the pigment layer, combines them with the optimized separation influence weights, and calculates the discrimination ratio. The discrimination ratio measures the degree of influence of the current candidate band on the separation of the ink layer and the pigment layer. The candidate band adjustment module determines whether the discrimination ratio is higher than the preset separation threshold. If it is, the current candidate band is output; otherwise, the candidate band is adjusted until the discrimination ratio is higher than the separation threshold, and the candidate band is output. The separation effect judgment module determines the center position of the band by constraining the light source with the output candidate band and the target sampling device, and performs separation of the ink layer and the pigment layer. The separation effect is judged based on the separation effect evaluation index and reference effect discrimination data, and the judgment result determines whether to adjust the optimized separation influence weights. The digital restoration and enhancement module separates the ink layer and pigment layer of the ancient Chinese painting based on the center position of the band, and then performs digital restoration and enhancement.

[0012] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0013] 1. The digital restoration and enhancement method for ancient Chinese paintings provided by this invention obtains a spectral sample set of ink layer and pigment layer and their corresponding spectral wavelengths, and calculates the separation difference degree. Simultaneously, it obtains the spectral reflectance data of the pigment layer spectral sample set and the initial separation influence weight of the pigment layer from a preset database. This allows the spectral difference between the ink layer and pigment layer to be quantitatively described, avoiding reliance on subjective experience or a single band for judgment. Next, a difference-wavelength variation curve is constructed based on the spectral wavelength and the corresponding separation difference degree. Candidate bands are then selected based on this curve, identifying bands with high discrimination potential from the continuous spectrum. This helps improve the targeting and effectiveness of subsequent band selection. Finally, the initial separation influence weight is adjusted according to the pigment aging degree reflected by the spectral reflectance data to obtain an optimized separation influence weight. This improves the method's adaptability to actual conditions such as pigment fading and deterioration in ancient paintings, enhancing the stability and accuracy of ink layer and pigment layer separation.

[0014] 2. By extracting features from each candidate band, the feature vectors of the ink layer and pigment layer are obtained. These feature vectors are then combined with the weights that influence the optimized separation to calculate the discrimination ratio. This allows the separation capability of different bands to be evaluated in a unified mathematical form. Next, it is determined whether the discrimination ratio is higher than the preset separation threshold. If so, the current candidate band is output. If not, the candidate bands are adjusted until the discrimination ratio is higher than the separation threshold, and then the candidate bands are output. This not only achieves automatic determination of the separation capability of candidate bands, but also the iterative screening mechanism can gradually approach the optimal band combination, improving the reliability of the final band selection.

[0015] 3. By constraining the light source of the output candidate bands and the target sampling device, the center position of the band is determined, and the ink layer and pigment layer are separated. The separation effect is judged based on the separation effect evaluation index and reference effect discrimination data. Based on the judgment result, it is determined whether to adjust the weight of the optimized separation. This helps to separate the ink layer and pigment layer under real imaging conditions. The separation effect evaluation index is used for verification, so that the band selection result can be tested in practical applications. It also improves the robustness of the method under different ancient books and different pigment combinations. Finally, the ink layer and pigment layer of the Chinese painting are separated according to the center position of the band, and then digital restoration and enhancement are performed. This makes the ink layer and pigment layer have good separability at the source data level, thereby improving the authenticity, clarity and stability of the digital restoration result of the Chinese painting and effectively solving the problem of decreased separation accuracy of ink layer and pigment layer in the existing technology.

[0016] 4. This invention calculates the optimized discrimination ratio for candidate bands and compares it with a separation threshold, achieving quantitative evaluation and automatic determination of the separation capability of the current candidate bands. If the optimized discrimination ratio is higher than the separation threshold, the corresponding optimized candidate band is output as a candidate band, which helps improve the efficiency of the band selection process. If the optimized discrimination ratio is not higher than the separation threshold, the difference between the optimized discrimination ratio and the separation threshold is used to query the second separation mapping table to obtain the fine-tuning ratio of the optimized candidate band. This not only achieves fine-tuning of candidate bands but also avoids large-scale, blind adjustments, thus improving the optimization process. The stability of the separation model is improved by optimizing the candidate bands and then performing a product operation on the optimized candidate bands by fine-tuning the ratio of the optimized candidate bands to obtain the second optimized candidate band. This improves the correlation between the optimized band and the original band and ensures the stable operation of the separation model. If the discrimination ratio of the second optimized candidate band is still lower than the separation threshold, a spectral analysis anomaly warning is issued and the preset staff is prompted to handle the anomaly, ensuring that the optimization process has a closed-loop verification mechanism. Otherwise, the second optimized candidate band is directly output as the candidate band, so that the system can still output effective bands under complex spectral conditions, thereby improving the success rate and stability of the separation process of ink layer and pigment layer in traditional Chinese painting.

[0017] 5. By weighted fusion of the deviation values ​​between each separation effect evaluation index and the corresponding reference separation effect evaluation data, a comprehensive separation effect evaluation quantity is obtained. This quantitatively characterizes the gap between the current separation result and the ideal separation effect, enabling the separation effect evaluation to simultaneously consider multiple dimensions such as spectral consistency, separation accuracy, and visual effect, thereby improving the comprehensiveness and reliability of the evaluation results. If the comprehensive separation effect evaluation quantity is greater than the preset separation effect evaluation value, the current ink layer and pigment layer are separated based on the current band center position, improving the system's operating efficiency and the real-time performance of the separation processing. If the comprehensive separation effect evaluation quantity is not greater than the separation effect evaluation value, the band center position is returned and a band center position re-analysis instruction is generated to prevent the band center position with insufficient separation effect from being directly used for final repair, reducing the risk of incorrect separation or repair. At the same time, the separation influence weight is finely adjusted based on the comprehensive separation effect evaluation quantity and the separation effect evaluation value, thereby realizing the dynamic change of the separation influence weight with the separation effect, avoiding the failure of fixed weights under complex pigment or aging conditions, and thus improving the accuracy and stability of subsequent band evaluation and separation decisions. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A flowchart illustrating the method for digital restoration and enhancement of ancient Chinese paintings and books provided in this application embodiment;

[0020] Figure 2 A schematic diagram illustrating the process of determining the discrimination ratio of optimized candidate bands provided in an embodiment of this application;

[0021] Figure 3 A schematic diagram of the separation effect judgment provided in the embodiments of this application;

[0022] Figure 4 A schematic diagram of the structure of the digital restoration and enhancement system for ancient Chinese paintings and books provided in this application embodiment. Detailed Implementation

[0023] The following provides explanations for some of the terms used in this application. It should be noted that these explanations are for the convenience of those skilled in the art and do not constitute a limitation on the scope of protection claimed in this application.

[0024] The embodiments of this application involve at least one, including one or more; where "multiple" means two or more. Furthermore, it should be understood that in the description of this specification, terms such as "first," "second," and "third" are used only for descriptive purposes and should not be construed as indicating relative importance or order. For example, "first device" and "second device" do not represent the degree of importance of the two or their order, but are merely for distinguishing what is described. In the embodiments of this application, "and / or" merely describes an association relationship, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone. Additionally, the character " / " in this document generally indicates that the preceding and following related objects have an "or" relationship.

[0025] The directional terms mentioned in the embodiments of this application, such as "up", "down", "left", "right", "inner", and "outer", are only for reference to the directions in the accompanying drawings. Therefore, the directional terms used are for better and clearer explanation and understanding of the embodiments of this application, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the embodiments of this application.

[0026] References to "one embodiment," "in some examples," or "some embodiments" as described in the embodiments of this application mean that one or more embodiments of this specification include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in some examples," "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0027] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0028] like Figure 1 The diagram shown is a flowchart illustrating the digital restoration and enhancement method for ancient Chinese paintings provided in this application embodiment. The method includes the following steps:

[0029] S1. Obtain the spectral sample set of the ink layer and the spectral sample set of the pigment layer and their corresponding spectral wavelengths, and calculate the separation difference to obtain the separation difference degree. At the same time, obtain the spectral reflectance data of the pigment layer spectral sample set, and obtain the initial separation influence weight of the pigment layer from the preset database. The initial separation influence weight characterizes the degree of separation contribution of the ink layer and the pigment layer at different spectral wavelengths. The separation difference degree is used to characterize the distance or similarity index of the separability of the ink layer spectrum and the pigment layer spectrum. The distance or similarity index includes at least one or more of the following: spectral angular distance, normalized Euclidean distance, and correlation coefficient difference.

[0030] Among them, the spectral wavelength is obtained through the calibration parameters of the multispectral acquisition device (multispectral or hyperspectral camera) or the spectral database; under the same spectral wavelength, the reflectance of the spectral samples of the ink layer and the pigment layer is statistically analyzed, and the separation difference is calculated based on the difference between the two. The separation difference is used to characterize the degree of distinguishability between the ink layer and the pigment layer under the spectral wavelength.

[0031] It should be explained that before designing methods and systems for the digital restoration and enhancement of traditional Chinese paintings and ancient books, technical professionals typically pre-construct a preset database to support the operation of various control strategies. This database integrates multiple key control parameters, including initial separation influence weights, separation thresholds, mapping data tables, separation mapping tables, reference separation effect evaluation data, separation effect evaluation values, separation effect intervals, first preset ratio, second preset ratio, third preset ratio, spectral consistency error threshold, separation error degree interval, and maximum fine-tuning ratio. The analytical logic and actual hardware environment of the methods and systems for the digital restoration and enhancement of traditional Chinese paintings and ancient books have all parameters pre-set by the technical professionals. This preset database provides crucial data support for subsequent data uploading, storage optimization, and even automated screening and judgment.

[0032] S2, a difference-wavelength variation curve is constructed based on the spectral wavelength and the corresponding separation difference degree. Candidate bands are selected based on this curve. Candidate bands represent the range of spectral wavelengths covered by the difference-wavelength variation curve where the separation difference degree is higher than the difference threshold.

[0033] It should be explained that the difference-wavelength variation curve uses spectral wavelength as the horizontal axis and separation difference as the vertical axis to describe the corresponding relationship between the two; the difference threshold is read from the preset database, which is generally preset by the staff based on historical data within a historical time period and existing work experience, and entered into the preset database. When using it, it can be directly read from the preset database.

[0034] S3, the initial separation influence weight is adjusted based on the pigment aging degree reflected by the spectral reflectance data to obtain the optimized separation influence weight.

[0035] It should be added that the specific process of obtaining the optimized separation influence weights is as follows:

[0036] First, the aging degree value, which reflects the aging degree of the pigment in the pigment layer, is obtained through spectral reflectance data. The spectral reflectance data includes spectral difference, characteristic peak intensity change rate, and spectral slope change.

[0037] Specifically, spectral difference The calculation expression is: , The average reflectance of the ink layer sample at this spectral wavelength. The average reflectance of the pigment layer sample at this spectral wavelength. Indicates spectral wavelength; rate of change of characteristic peak intensity Among them, the current pigment peak intensity Reference peak intensity , The characteristic peak wavelength, The spectral reflectance curve of the pigment layer; the change in spectral slope. , For band intervals, the data source is still from .

[0038] Next, the aging degree value is input into a preset mapping data table for comparison to obtain the corresponding aging degree correction amount.

[0039] The aging degree value is input into the mapping data table to obtain the corresponding aging degree correction amount. This data sequence is used to fit the mapping relationship between the aging degree value and the aging degree correction amount. The construction method is as follows: in the initial data table built based on the gradient boosting regression algorithm, the aging degree values ​​collected in the historical time period and the aging degree correction amount set according to the empirical rules are selected as training samples. The model is trained based on the XGBoost framework with the least square error as the objective function, and finally the trained mapping data table is obtained.

[0040] Finally, the optimized separation influence weight is obtained by multiplying the aging correction amount with the separation influence weight.

[0041] By incorporating the spectral reflectance data of the pigment layer into the separation process, quantitative compensation and adaptive adjustment of pigment aging factors are achieved. Specifically, by comprehensively considering spectral differences, characteristic peak intensity change rate, and spectral slope change, an aging degree value is constructed, allowing for an objective description of the spectral changes of pigments under different degradation states. Then, a corresponding aging degree correction value is obtained through a preset mapping data table and used to correct the separation influence weight, thereby enabling the separation parameters to be dynamically adjusted according to the pigment aging degree. This method effectively avoids misjudgments caused by pigment fading or spectral drift in traditional separation processes, improves the accuracy and robustness of ink layer and pigment layer separation, and enhances the adaptability of the method in complex aging scenarios of ancient Chinese paintings.

[0042] S4. Feature extraction is performed on each candidate band to obtain the feature vectors of the ink layer and the pigment layer. These feature vectors are then combined with the optimized separation influence weights to calculate the discriminant ratio. The discriminant ratio is used to measure the degree of influence of the current candidate band on the separation of the ink layer and the pigment layer.

[0043] Specifically, spectral statistical features, spectral morphological features, and band difference features of the ink layer and pigment layer are extracted for each candidate band to construct feature vectors for the ink layer and pigment layer. The features include, but are not limited to, spectral mean, variance, local slope, curvature, and spectral difference values ​​of the ink layer and pigment layer in the same band. Feature extraction methods include, but are not limited to, single-band statistical features, multi-band joint statistical features, PCA (Principal Component Analysis), and LDA (Linear Discriminant Analysis).

[0044] Specifically, the discriminant ratio , The discrimination ratio between the ink layer and the pigment layer. and These represent the characteristic mean vectors of the ink layer and pigment layer in the candidate bands, respectively. and These represent the within-class variance of the corresponding features. This indicates the weights that influence the optimization separation.

[0045] S5. Determine whether the discrimination ratio is higher than the preset separation threshold. If yes, output the current candidate band. If no, adjust the candidate band until the discrimination ratio is higher than the separation threshold and output the candidate band. The separation threshold is read from the preset database and is generally preset by professionals based on experience rules.

[0046] The specific process for adjusting candidate bands to obtain optimized candidate bands is as follows:

[0047] The discriminant ratio and the separation threshold are calculated by difference, and then the ratio is calculated with the separation threshold to obtain the corresponding discriminant separation rate.

[0048] The separation rate is input into a preset separation mapping table for matching to obtain the corresponding candidate band adjustment ratio.

[0049] It should be added that by inputting the judgment separation rate into the separation mapping table, the corresponding candidate band adjustment ratio can be obtained. This dataset is used to fit the mapping relationship between the judgment separation rate and the candidate band adjustment ratio. The construction method is as follows: in the initial data table built based on the linear regression algorithm, the judgment separation rate collected in the historical time period and the candidate band adjustment ratio set according to the empirical rules are selected as training samples. The least squares criterion is used and the model is trained based on the statsmodels framework to finally obtain the trained separation mapping table.

[0050] The optimized candidate band is obtained by multiplying the candidate band adjustment ratio with the corresponding candidate band.

[0051] like Figure 2 The diagram shows a flowchart of the discrimination ratio determination process for optimized candidate bands provided in this application embodiment. The specific logic is as follows: The optimized discrimination ratio is calculated for each optimized candidate band. The optimized discrimination ratio is then compared with a separation threshold. If the optimized discrimination ratio is higher than the separation threshold, the corresponding optimized candidate band is output as a candidate band. If the optimized discrimination ratio is not higher than the separation threshold, the difference between the optimized discrimination ratio and the separation threshold is used to query the second separation mapping table to obtain the fine-tuning ratio of the optimized candidate band. The optimized candidate band is then multiplied using the fine-tuning ratio to obtain a second optimized candidate band. If the discrimination ratio of the second optimized candidate band is still lower than the separation threshold, a spectral analysis anomaly warning is issued, and a preset staff member is prompted to handle the anomaly. Otherwise, the second optimized candidate band is directly output as a candidate band. This process helps to quickly determine whether the candidate band meets the separation requirements of the ink layer and the pigment layer, thereby improving the separation accuracy of the ink layer and the pigment layer.

[0052] As a further embodiment, obtaining optimized candidate bands also includes:

[0053] The first step is to calculate the optimized discrimination ratio for the optimized candidate bands, and then compare the optimized discrimination ratio with the separation threshold. This means that the feature mean vector and intra-class variance in the optimized candidate bands are substituted back into the discrimination ratio calculation expression to obtain the optimized discrimination ratio.

[0054] The second step is to output the corresponding optimized candidate band as a candidate band if the optimized discrimination ratio is higher than the separation threshold. If the optimized discrimination ratio is not higher than the separation threshold, the optimized candidate band fine-tuning ratio is obtained by querying the second separation mapping table based on the difference between the optimized discrimination ratio and the separation threshold.

[0055] Similarly, the difference between the optimized discrimination ratio and the separation threshold is denoted as the difference in discrimination ratio. By inputting the difference in discrimination ratio into the second separation mapping table, the corresponding optimized candidate band fine-tuning ratio can be obtained. This data table is used to fit the mapping relationship between the difference in discrimination ratio and the optimized candidate band fine-tuning ratio. Its construction method is as follows: In the initial dataset constructed based on the linear regression algorithm, the difference in discrimination ratio collected in the historical time period and the optimized candidate band fine-tuning ratio set according to empirical rules are selected as training samples. The least squares criterion is used and the model is trained based on the statsmodels framework to finally obtain the trained second separation mapping table.

[0056] The third step is to perform a product operation on the optimized candidate bands by optimizing the fine-tuning ratio of the candidate bands to obtain the second optimized candidate bands;

[0057] Fourth step: If the discrimination ratio of the second optimized candidate band is still lower than the separation threshold, a spectral analysis anomaly warning will be issued and the preset staff will be prompted to handle the anomaly; otherwise, the second optimized candidate band will be directly output as the candidate band.

[0058] By introducing a hierarchical judgment and adaptive fine-tuning mechanism based on the discriminant ratio threshold, the candidate band screening process is made more refined and safer. Furthermore, by comparing the discriminant ratio of optimized candidate bands with the separation threshold, it is possible to quickly determine whether they meet the separation requirements of the ink layer and pigment layer. When the discrimination capability is insufficient, the fine-tuning ratio is obtained from the preset second separation mapping table based on the difference between the discriminant ratio and the threshold, and the candidate bands are quantitatively corrected, thereby avoiding the direct rejection of potentially effective bands. At the same time, by setting up a secondary discrimination and anomaly warning mechanism, manual intervention is promptly prompted when the separation capability is continuously insufficient, effectively preventing abnormal data or extreme aging conditions from affecting the separation results.

[0059] S6 determines the center position of the band by constraining the light source of the output candidate bands and the target sampling device (such as a multispectral camera or hyperspectral camera), and separates the ink layer and pigment layer. The separation effect is judged according to the separation effect evaluation index and the reference effect discrimination data, and the weight of the effect on the optimization separation is determined based on the judgment result.

[0060] In some examples, the process of constraining a light source first examines the device parameters: the range of bands that can be covered, the center wavelength of the available filters, and the bandwidth limitations; then, it selects the center wavelength from the candidate bands that is closest to the device and that the light source has sufficient energy output in that candidate band.

[0061] For example, the theoretically optimal band center position is 780nm, while the target sampling device has available band center positions of 760nm and 800nm. Based on the above selection criteria, 800nm ​​is selected as the corresponding band center position.

[0062] like Figure 3The diagram shows a flowchart of the separation effect judgment provided in this application embodiment. The specific process is as follows: Obtain separation effect evaluation indicators; compare each separation effect evaluation indicator with reference separation effect evaluation data; if each separation effect evaluation indicator meets the judgment conditions corresponding to its reference separation effect evaluation data, output the corresponding band center position for separation; if none of the separation effect evaluation indicators meet the judgment conditions corresponding to the corresponding reference separation effect evaluation data, return a band center position error and prompt for re-filtering the band center position; if not all separation effect evaluation indicators meet the judgment conditions corresponding to the corresponding reference separation effect evaluation data, obtain the deviation values ​​between each separation effect evaluation indicator and the corresponding reference separation effect evaluation data, perform weighted fusion, and obtain a comprehensive separation effect evaluation value; if the comprehensive separation effect evaluation value is greater than the separation effect evaluation value, then based on the current band center position... The process involves separating the current ink layer from the pigment layer. If the overall separation effect assessment is not greater than the separation effect assessment value, the process returns to the band center position and generates a band center position reanalysis command. Simultaneously, the overall separation effect assessment and the separation effect assessment value are subtracted to obtain the corresponding separation effect difference, which is then matched with a preset separation effect interval. If the separation effect difference belongs to the first separation effect interval, the weight of the optimized separation influence is multiplied according to the first preset ratio. If the separation effect difference belongs to the second separation effect interval, the weight of the optimized separation influence is multiplied according to the second preset ratio. Otherwise, the weight of the optimized separation influence is multiplied according to the third preset ratio. Based on the above process, not only is the reliability of the separation result assessment improved and the risk of misjudgment reduced, but the separation parameters can also adapt to complex actual conditions such as pigment aging and fading, which helps to improve the robustness of the separation process.

[0063] Furthermore, the specific process for determining whether to adjust the weights affecting the optimization separation based on the judgment results is as follows:

[0064] The separation performance evaluation metrics include precision, recall, and F1 score. If labeled data exists, a confusion matrix is ​​constructed by comparing the classification results (determined as ink layer or pigment layer). The quantities of the following four categories are counted at the pixel, region, or sample level:

[0065] name meaning TP (True Positive) It is actually an ink layer, and has been correctly separated into ink layers. FP (False Positive) It is actually a pigment layer, but it has been mistakenly separated into an ink layer. FN (False Negative) It is actually an ink layer, but it has been mistakenly separated into a pigment layer. TN (True Negative) It is actually a pigment layer, and has been correctly separated into pigment layers.

[0066] Based on the above statistical results, accuracy Recall rate F1 value .

[0067] Each separation effect evaluation index is compared with the preset reference separation effect evaluation data, resulting in the following three scenarios:

[0068] The first method: If each separation effect evaluation index meets the judgment conditions corresponding to its reference separation effect evaluation data, then the corresponding band center position is output for separation; the reference separation effect evaluation data includes the minimum precision, the minimum recall, and the minimum F1 value, all of which are extracted from the preset database and are also preset by the preset staff based on historical data and experience rules.

[0069] The criteria for determining F1 score are: precision greater than the minimum precision, recall greater than the minimum recall, and F1 score higher than the minimum F1 score.

[0070] The second approach: If none of the separation effect evaluation indicators meet the judgment conditions corresponding to the reference separation effect evaluation data, then an error in the band center position will be returned and a prompt will be made to re-filter the band center position.

[0071] The third approach is to conduct a comprehensive analysis of the separation effect if not all the separation effect evaluation indicators meet the corresponding judgment conditions of the reference separation effect evaluation data.

[0072] By introducing a multi-dimensional separation effect evaluation and grading judgment mechanism, objective verification and reliable control of the separation results of the ink layer and pigment layer are achieved. Simultaneously, by acquiring precision, recall, and F1 score, and comparing them with preset reference separation effect evaluation data, the separation effect of the current band center position can be comprehensively evaluated from multiple perspectives, including accuracy, completeness, and overall performance. When all indicators meet the judgment conditions, the validity of the band center position can be directly confirmed and used for separation processing. When none of the indicators meet the conditions, an error message is promptly returned and re-screening is guided to prevent invalid bands from entering subsequent processes. Furthermore, when some indicators are met, potential optimization space is further explored through comprehensive analysis of the separation effect. This scheme improves the comprehensiveness and reliability of the separation result evaluation, reduces the risk of misjudgment, and enhances the stability and practicality of the system in complex scenarios.

[0073] It should be added that the specific content of the comprehensive analysis of separation effect is as follows: obtain the deviation values ​​between each separation effect evaluation index and the corresponding reference separation effect evaluation data, and then perform weighted fusion to obtain the comprehensive evaluation quantity of separation effect.

[0074] Specifically, after calculating the difference between each separation effect evaluation index and the corresponding reference separation effect evaluation data, the data is normalized, multiplied by the weights extracted from the preset data, and then summed to obtain the corresponding comprehensive evaluation value of separation effect.

[0075] If the overall evaluation of the separation effect is greater than the preset evaluation value of the separation effect, then the current ink layer and pigment layer will be separated based on the current center position of the current band.

[0076] If the comprehensive evaluation of the separation effect is not greater than the evaluation value of the separation effect, the band center position is returned and a re-analysis instruction for the band center position is generated. At the same time, the separation effect weight is fine-tuned based on the comprehensive evaluation of the separation effect and the evaluation value of the separation effect.

[0077] The specific methods for optimizing the separation of influence weights and fine-tuning them are as follows:

[0078] The difference between the comprehensive evaluation of the separation effect and the evaluation value of the separation effect is calculated to obtain the corresponding separation effect difference, and then matched with the preset separation effect interval. The separation effect interval includes the first separation effect interval, the second separation effect interval, and the third separation effect interval, which represent the separation effect in ascending order. The separation effect interval is a preset data range for the preset staff, which is usually set according to historical data and empirical rules.

[0079] If the poor separation effect falls within the first separation effect range, then the weight of the degree of influence of the optimized separation is multiplied according to the first preset ratio. If the poor separation effect falls within the second separation effect range, then the weight of the degree of influence of the optimized separation is multiplied according to the second preset ratio. Otherwise, the weight of the degree of influence of the optimized separation is multiplied according to the third preset ratio.

[0080] S7 separates the ink and pigment layers of ancient Chinese paintings based on the center position of the band, and then performs digital restoration and enhancement.

[0081] In this embodiment, a band-adaptive screening and optimization mechanism based on spectral difference and pigment aging characteristics is introduced to achieve high-precision separation of ink and pigment layers in ancient Chinese paintings. First, by acquiring the spectral sample sets of ink and pigment layers and their corresponding spectral wavelengths, and calculating the separation difference between them, a difference-wavelength variation curve is constructed. This intuitively reflects the separability of ink and pigment layers in the spectral dimension, providing clear data basis for the selection of candidate bands and avoiding the problem of traditional band selection relying on experience or a single indicator. Second, by combining pigment layer spectral reflectance data to evaluate the degree of pigment aging, and on this basis, the initial separation influence weight is dynamically adjusted, enabling the separation parameters to adapt to complex actual conditions such as pigment aging and fading, significantly improving the robustness and adaptability of the separation process.

[0082] Furthermore, this invention extracts feature vectors of the ink layer and pigment layer from the candidate bands respectively, and combines the feature vectors with the optimized separation influence weights to calculate the discriminant ratio. This quantifies the influence of each candidate band on the separation of the ink layer and pigment layer from the dual perspectives of feature distribution and weight adjustment. Moreover, by comparing the discriminant ratio with a preset separation threshold, automatic screening and iterative optimization of candidate bands are achieved, effectively preventing bands with insufficient separation effect from entering the subsequent processing flow. Finally, the center position of the band is determined by constraining the light source between the output candidate bands and the target sampling device, and a feedback adjustment mechanism is formed by combining the separation effect evaluation index. This allows the separation influence weights to be further corrected according to the actual separation effect, thereby constructing a closed-loop, adaptive separation optimization process.

[0083] When spectral samples lack precise labeling, the separation effect is judged based on the separation effect evaluation index and reference effect discrimination data, and this also includes:

[0084] The spectral consistency error is obtained and compared with a preset spectral consistency error threshold. , The spectrum of the separated ink layer, Using the reference ink spectrum, the spectral consistency error threshold is the maximum error value that limits the fluctuation of the spectral consistency error.

[0085] If the spectral consistency error is less than the spectral consistency error threshold, the ink layer and pigment layer are separated according to the center position of the corresponding band.

[0086] If the spectral consistency error is not less than the spectral consistency error threshold, then the difference between the spectral consistency error and the spectral consistency error threshold is recorded as the difference in spectral consistency, and the weight of the degree of influence of the optimization separation is adjusted based on the difference in spectral consistency.

[0087] By introducing spectral consistency error as a supplementary constraint in the separation effect judgment process, the reliability and stability of the ink layer and pigment layer separation results are further improved. By comparing the spectral consistency error with a preset threshold, the rationality of the separation result at the current band center position can be verified from the perspective of spectral matching. When the consistency error is small, the separation operation is directly executed, effectively ensuring the degree of matching between the separation result and the true spectral characteristics. When the consistency error exceeds the threshold, the weight of the degree of influence of the optimized separation is adjusted based on the error difference, so that the separation parameters can adaptively correct for spectral mismatch. This mechanism avoids the bias that may be caused by relying on a single separation effect evaluation index, enhances the system's adaptability to complex spectral changes and abnormal situations, and improves the overall separation accuracy and robustness.

[0088] The specific details of adjusting the weighting of the impact of spectral consistency differences on the degree of separation optimization are as follows:

[0089] The difference in spectral consistency is compared with a preset separation error range, which includes a first separation error range, a second separation error range, a third separation error range, and a fourth separation error range, with the degree of separation error represented decreasing sequentially.

[0090] If the difference in spectral consistency falls within the first separation error range, the weight of the degree of influence of the optimized separation is adjusted at the first level. If the difference in spectral consistency falls within the second separation error range, the weight of the degree of influence of the optimized separation is adjusted at the second level. If the difference in spectral consistency falls within the third separation error range, the weight of the degree of influence of the optimized separation is adjusted at the third level. Otherwise, the weight of the degree of influence of the optimized separation is adjusted at the fourth level.

[0091] Level 1 adjustment means multiplying the preset maximum fine-tuning ratio with the weight of the degree of influence of the optimization separation; Level 2 adjustment means multiplying the maximum fine-tuning ratio and its level 2 ratio with the weight of the degree of influence of the optimization separation; Level 3 adjustment means multiplying the maximum fine-tuning ratio and its level 3 ratio with the weight of the degree of influence of the optimization separation; Level 4 adjustment means multiplying the maximum fine-tuning ratio and its level 4 ratio with the weight of the degree of influence of the optimization separation.

[0092] The aforementioned separation error range and maximum fine-tuning ratio are all preset values, usually set by professionals based on historical data, work experience, and industry rules. In addition, the secondary, tertiary, and quaternary ratios are usually set to 0.75, 0.5, and 0.25, respectively.

[0093] like Figure 4 The diagram shown is a structural schematic of the digital restoration and enhancement system for ancient Chinese paintings and books provided in this application embodiment, including:

[0094] The digital data acquisition module is used to acquire the spectral sample sets of the ink layer and the pigment layer and their corresponding spectral wavelengths, and to calculate the separation difference degree. At the same time, it acquires the spectral reflectance data of the pigment layer spectral sample set and obtains the initial separation influence weight of the pigment layer from the preset database.

[0095] The digital data processing module is used to construct a difference-wavelength variation curve based on the spectral wavelength and the corresponding separation difference degree, and to select candidate bands accordingly. The candidate bands represent the range of spectral wavelengths covered by the difference-wavelength variation curve where the separation difference degree is higher than the difference threshold.

[0096] The pigment layer weight adjustment module is used to adjust the initial separation influence weight based on the pigment aging degree reflected by the spectral reflectance data to obtain the optimized separation influence weight.

[0097] The candidate band separation effect evaluation module is used to extract features from each candidate band to obtain the ink layer feature vector and the pigment layer feature vector. These features are then combined with the optimized separation influence weight to calculate the discrimination ratio. The discrimination ratio is used to measure the degree of influence of the current candidate band on the separation of the ink layer and the pigment layer.

[0098] The candidate band adjustment module is used to determine whether the discrimination ratio is higher than the preset separation threshold. If yes, the current candidate band is output; otherwise, the candidate band is adjusted until the discrimination ratio is higher than the separation threshold, and the candidate band is output.

[0099] The separation effect judgment module is used to determine the center position of the band by constraining the light source between the output candidate band and the target sampling device, and to separate the ink layer and pigment layer. It judges the separation effect based on the separation effect evaluation index and reference effect discrimination data, and determines whether to adjust the weight of the separation effect based on the judgment result.

[0100] The digital restoration and enhancement module is used to separate the ink layer and pigment layer of ancient Chinese paintings based on the center position of the band, and then perform digital restoration and enhancement.

[0101] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to the embodiments of the present invention is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium (e.g., a solid-state drive (SSD)). Where there is no conflict, the solutions in the above embodiments can be combined.

[0102] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0103] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to this application. It should be understood that each block of the flowchart illustrations and / or block diagrams, as well as combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0104] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0105] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0106] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for digital restoration and enhancement of ancient Chinese paintings and books, characterized in that, Includes the following steps: The process involves acquiring a spectral sample set of the ink layer and a spectral sample set of the pigment layer, along with their corresponding spectral wavelengths, and calculating the separation difference to obtain the separation difference degree. Simultaneously, the spectral reflectance data of the pigment layer spectral sample set is acquired, as well as the initial separation influence weight of the pigment layer. The initial separation influence weight characterizes the degree of separation contribution of the ink layer and the pigment layer at different spectral wavelengths. The separation difference degree is used to characterize the distance or similarity index of the separability of the ink layer spectrum and the pigment layer spectrum. The distance or similarity index includes at least one or more of the following: spectral angular distance, normalized Euclidean distance, and correlation coefficient difference. Based on the spectral wavelength and the corresponding separation difference degree, a difference-wavelength variation curve is constructed, and candidate bands are selected accordingly. The candidate bands represent the spectral wavelength range covered by the difference-wavelength variation curve where the separation difference degree is higher than the difference threshold. The initial separation influence weights are adjusted based on the pigment aging degree reflected by the spectral reflectance data to obtain the optimized separation influence weights; Feature extraction is performed on each candidate band to obtain the ink layer feature vector and the pigment layer feature vector. These are then combined with the optimized separation influence weight to calculate the discrimination ratio. The discrimination ratio is used to measure the degree of influence of the current candidate band on the separation of the ink layer and the pigment layer. Determine whether the discrimination ratio is higher than a preset separation threshold. If yes, output the current candidate band. If no, adjust the candidate band until the discrimination ratio is higher than the separation threshold, and output the candidate band. The center position of the band is determined by constraining the light source with the output candidate band and the target sampling device, and the ink layer and pigment layer are separated. The separation effect is judged according to the separation effect evaluation index and the reference effect discrimination data, and the weight of the effect on the optimization separation is determined based on the judgment result. Based on the center position of the band, the ink layer and pigment layer of ancient Chinese paintings are separated, and then digital restoration and enhancement are carried out.

2. The method for digital restoration and enhancement of ancient Chinese paintings and books as described in claim 1, characterized in that: The specific process for obtaining the optimized separation influence weights is as follows: The aging degree value, which reflects the aging degree of the pigment in the pigment layer, is obtained by using spectral reflectance data. The spectral reflectance data includes spectral difference, characteristic peak intensity change rate, and spectral slope change. The aging degree value is input into a preset mapping data table for comparison to obtain the corresponding aging degree correction amount; The optimized separation influence weight is obtained by multiplying the aging correction amount with the separation influence weight.

3. The method for digital restoration and enhancement of ancient Chinese paintings and books as described in claim 1, characterized in that: The specific process for adjusting the candidate bands until the discrimination ratio is not higher than the separation threshold is as follows: The discriminant ratio and the separation threshold are calculated by difference, and then the ratio is calculated with the separation threshold to obtain the corresponding discriminant separation rate. The separation rate is input into a preset separation mapping table for matching to obtain the corresponding candidate band adjustment ratio; The optimized candidate band is obtained by multiplying the candidate band adjustment ratio with the corresponding candidate band.

4. The method for digital restoration and enhancement of ancient Chinese paintings and books as described in claim 3, characterized in that: The process of obtaining optimized candidate bands also includes: The optimized discrimination ratio is calculated for the optimized candidate bands, and then compared with the separation threshold. If the optimized discrimination ratio is higher than the separation threshold, the corresponding optimized candidate band will be output as a candidate band. If the optimized discrimination ratio is not higher than the separation threshold, the optimized candidate band fine-tuning ratio is obtained by querying the second separation mapping table based on the difference between the optimized discrimination ratio and the separation threshold. By optimizing the candidate band fine-tuning ratio, the optimized candidate bands are further multiplied to obtain the second optimized candidate band; If the discrimination ratio of the second optimized candidate band is still lower than the separation threshold, an abnormal spectral analysis warning will be issued and the preset staff will be prompted to handle the abnormality; otherwise, the second optimized candidate band will be directly output as the candidate band.

5. The method for digital restoration and enhancement of ancient Chinese paintings and books as described in claim 1, characterized in that: The specific process for determining whether to adjust the weights affecting the optimization separation based on the judgment result is as follows: Obtain separation performance evaluation metrics, including precision, recall, and F1 score; Each separation performance evaluation index was compared with the preset reference separation performance evaluation data: If each separation effect evaluation index meets the judgment condition corresponding to its reference separation effect evaluation data, then the center position of the corresponding band will be output for separation. If none of the separation effect evaluation indicators meet the judgment conditions corresponding to the reference separation effect evaluation data, an error in the band center position will be returned and a prompt will be made to re-filter the band center position. If not all separation effect evaluation indicators meet the judgment conditions corresponding to the reference separation effect evaluation data, a comprehensive analysis of the separation effect will be conducted.

6. The method for digital restoration and enhancement of ancient Chinese paintings and books as described in claim 5, characterized in that: The specific details of the comprehensive analysis of the separation effect are as follows: The deviation values ​​between each separation effect evaluation index and the corresponding reference separation effect evaluation data are obtained and weighted and fused to obtain the comprehensive evaluation value of separation effect. If the overall evaluation value of the separation effect is greater than the preset evaluation value of the separation effect, then the current ink layer and pigment layer will be separated based on the current center position of the current band. If the comprehensive evaluation of the separation effect is not greater than the evaluation value of the separation effect, the band center position is returned and a re-analysis instruction for the band center position is generated. At the same time, the separation effect weight is fine-tuned based on the comprehensive evaluation of the separation effect and the evaluation value of the separation effect.

7. The method for digital restoration and enhancement of ancient Chinese paintings and books as described in claim 6, characterized in that: The specific method for optimizing and separating the influence weights for fine-tuning is as follows: The difference between the comprehensive evaluation value of the separation effect and the evaluation value of the separation effect is calculated to obtain the corresponding separation effect difference, and then matched with the preset separation effect interval. The separation effect interval includes a first separation effect interval, a second separation effect interval, and a third separation effect interval, and the separation effect represented by them increases sequentially. If the poor separation effect falls within the first separation effect range, then the weight of the degree of influence of the optimized separation is multiplied according to the first preset ratio. If the poor separation effect falls within the second separation effect range, then the weight of the degree of influence of the optimized separation is multiplied according to the second preset ratio. Otherwise, the weight of the degree of influence of the optimized separation is multiplied according to the third preset ratio.

8. The method for digital restoration and enhancement of ancient Chinese paintings and books as described in claim 1, characterized in that: The step of judging the separation effect based on the separation effect evaluation index and the reference effect discrimination data also includes: Obtain the spectral consistency error and compare it with a preset spectral consistency error threshold: If the spectral consistency error is less than the spectral consistency error threshold, the ink layer and pigment layer are separated according to the center position of the corresponding band. If the spectral consistency error is not less than the spectral consistency error threshold, then the difference between the spectral consistency error and the spectral consistency error threshold is recorded as the difference in spectral consistency, and the weight of the degree of influence of the optimization separation is adjusted based on the difference in spectral consistency.

9. The method for digital restoration and enhancement of ancient Chinese paintings and books as described in claim 8, characterized in that: The specific details of adjusting the weighting of the impact of spectral consistency difference on the optimized separation are as follows: The difference in spectral consistency is compared with a preset separation error range, which includes a first separation error range, a second separation error range, a third separation error range, and a fourth separation error range, with the degree of separation error represented decreasing sequentially. If the difference in spectral consistency falls within the first separation error range, the weight of the degree of influence of the optimized separation is adjusted at the first level; if the difference in spectral consistency falls within the second separation error range, the weight of the degree of influence of the optimized separation is adjusted at the second level; if the difference in spectral consistency falls within the third separation error range, the weight of the degree of influence of the optimized separation is adjusted at the third level; otherwise, the weight of the degree of influence of the optimized separation is adjusted at the fourth level. The first-level adjustment means multiplying the preset maximum fine-tuning ratio with the weight of the degree of influence of the optimization separation; the second-level adjustment means multiplying the maximum fine-tuning ratio and its second-level ratio with the weight of the degree of influence of the optimization separation; the third-level adjustment means multiplying the maximum fine-tuning ratio and its third-level ratio with the weight of the degree of influence of the optimization separation; and the fourth-level adjustment means multiplying the maximum fine-tuning ratio and its fourth-level ratio with the weight of the degree of influence of the optimization separation.

10. A digital restoration and enhancement system for ancient Chinese paintings and books, characterized in that: include: The digital data acquisition module is used to acquire the spectral sample set of ink layer and pigment layer and their corresponding spectral wavelengths, and to perform difference calculations to obtain the separation difference degree. At the same time, it acquires the spectral reflectance data of the pigment layer spectral sample set and obtains the initial separation influence weight of the pigment layer from the preset database. The digital data processing module is used to construct a difference-wavelength variation curve based on the spectral wavelength and the corresponding separation difference degree, and to screen candidate bands accordingly. The candidate bands represent the range of spectral wavelengths in the difference-wavelength variation curve whose separation difference degree is higher than the difference threshold. The pigment layer weight adjustment module is used to adjust the initial separation influence weight according to the pigment aging degree reflected by the spectral reflectance data to obtain the optimized separation influence weight; The candidate band separation effect evaluation module is used to extract features from each candidate band to obtain the ink layer feature vector and the pigment layer feature vector, and combine them with the optimized separation influence weight to calculate the discrimination ratio. The discrimination ratio is used to measure the degree of influence of the current candidate band on the separation of the ink layer and the pigment layer. The candidate band adjustment module is used to determine whether the discrimination ratio is higher than a preset separation threshold. If yes, the current candidate band is output; if no, the candidate band is adjusted until the discrimination ratio is higher than the separation threshold, and the candidate band is output. The separation effect judgment module is used to determine the center position of the band by constraining the light source with the output candidate band and the target sampling device, and to separate the ink layer and pigment layer. It judges the separation effect based on the separation effect evaluation index and reference effect discrimination data, and determines whether to adjust the weight of the optimized separation effect based on the judgment result. The digital restoration and enhancement module is used to separate the ink layer and pigment layer of ancient Chinese paintings based on the center position of the band, and then perform digital restoration and enhancement.

Images